Machine Learning for

Business Analytics

Machine learning is an integral tool in a business analyst’s arsenal because

the rate at which data is being generated from different sources is increasing
and working on complex unstructured data is becoming inevitable. Data
collection, data cleaning, and data mining are rapidly becoming more
diffcult to analyze than just importing information from a primary or
secondary source. The machine learning model plays a crucial role in
predicting the future performance and results of a company. In real time,
data collection and data wrangling are the important steps in deploying
the models. Analytics is a tool for visualizing and steering data and
statistics. Business analysts can work with different data sets—choosing an
appropriate machine learning model results in correct analyzing, forecasting
the future, and making informed decisions. The global machine learning
market was valued at $1.58 billion in 2017 and is expected to reach $20.83
billion in 2024—growing at a CAGR of 44.06% between 2017 and 2024. The
authors have compiled important knowledge on machine learning real-time
applications in business analytics. This book enables readers to get broad
knowledge in the feld of machine learning models and to carry out their
future research work. The future trends of machine learning for business
analytics are explained with real case studies. Essentially, this book acts as a
guide to all business analysts. The authors blend the basics of data analytics
and machine learning and extend its application to business analytics.
This book acts as a superb introduction and covers the applications and
implications of machine learning. The authors provide frst-hand experience
of the applications of machine learning for business analytics in the section
on real-time analysis. Case studies put the theory into practice so that you
may receive hands-on experience with machine learning and data analytics.
This book is a valuable source for practitioners, industrialists, technologists,
and researchers.
Machine Learning for
Business Analytics
Real-Time Data Analysis
for Decision-Making

Edited by
Hemachandran K., Sayantan Khanra,
Raul V. Rodriguez & Juan R. Jaramillo
First published 2023
by Routledge
605 Third Avenue, New York, NY 10158

and by Routledge
4 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN

Routledge is an imprint of the Taylor & Francis Group, an informa business

© 2023 selection and editorial matter, Hemachandran K., Sayantan Khanra,

Raul V. Rodriguez & Juan R. Jaramillo; individual chapters, the contributors

The right of Hemachandran K., Sayantan Khanra, Raul V. Rodriguez & Juan
R. Jaramillo to be identifed as the authors of the editorial material, and of
the authors for their individual chapters, has been asserted in accordance
with sections 77 and 78 of the Copyright, Designs and Patents Act 1988.

All rights reserved. No part of this book may be reprinted or reproduced or

utilised in any form or by any electronic, mechanical, or other means, now
known or hereafter invented, including photocopying and recording, or in
any information storage or retrieval system, without permission in writing
from the publishers.

Trademark notice: Product or corporate names may be trademarks or

registered trademarks, and are used only for identifcation and explanation
without intent to infringe.

Library of Congress Cataloging-in-Publication Data

A catalog record for this title has been requested

ISBN: 978-1-032-07281-4 (hbk)

ISBN: 978-1-032-07277-7 (pbk)
ISBN: 978-1-003-20631-6 (ebk)

DOI: 10.4324/9781003206316

Typeset in Garamond
by Apex CoVantage, LLC
Contents

Preface .......................................................................................................vii
Editors’ Biographies.....................................................................................ix
List of Contributors ...................................................................................xiii

1 Introduction to Machine Learning for Data Analytics................... 1

L. K. INDUMATHI, ABDUL RAIS, AND JUVAIRIA BEGUM
2 Role of Machine Learning in Promoting Sustainability................17
MUNEZA KAGZI
3 Addressing the Utilization of Popular Regression Models in
Business Applications..................................................................... 29
MEGANATHAN KUMAR SATHEESH AND
KORUPALLI V. RAJESH KUMAR
4 Chatbots: Their Uses and Impact in the Hospitality Sector..........45
PRINCY SERA RAJAN, DARSANA S. BABU, AND
SAMEENA M. H.
5 Traversing through the Use of Robotics in the Medical
Industry: Outlining Emerging Trends and Perspectives for
Future Growth .................................................................................57
GAURAV NAGPAL, KSHITIZ SINHA, HIMANSHU SETH,
AND NAMITA RUPAREL
6 Integration of AI in Insurance and Healthcare:
What Does It Mean? .........................................................................73
A. KANNAN, B. JUSTUS RABI, AND M. ANAND
7 Artifcial Intelligence in Agriculture: A Review ........................... 87
HARSHITHA SIRINENI, THAKUR SANTOSH,
AND S. DEEPAJOTHI

v
vi ◾ Machine Learning for Business Analytics

8 Machine Learning and Artifcial Intelligence-Based

Tools in Digital Marketing: An Integrated Approach ................... 97
PREETHA MARY GEORGE, SANJEEV GANGULY, AND
VENKAT REDDY YASA
9 Application of Artifcial Intelligence in Market Knowledge
and B2B Marketing Co-creation ....................................................107
H. RAGHUPATHI, DEBDUTTA CHOUDHURY, AND
CYNTHIA JABBOUR SFEIR
10 A Systematic Literature Review of the Impact of Artifcial
Intelligence on Customer Experience .......................................... 117
M. A. SIKANDAR, PRAVEEN KUMAR MUNARI, AND
MEGHRAJ ARLI
11 The Impact of Artifcial Intelligence on Customer
Experience and the Purchasing Process......................................129
LAXMI SHAW, MEGHA MANKAL, AND
CHINNAPANI KIRAN KUMAR
12 Application of Artifcial Intelligence in Banking:
A Review .........................................................................................145
SYED HASAN JAFAR, VIPLAV DHANDHUKIA, AND
BIJAY KUMAR G.
13 Digital Ethics: Toward a Socially Preferable Development
of AI Systems..................................................................................153
C. GUZMÁN-VELÁSQUEZ AND J. G. LALINDE-PULIDO

Index........................................................................................................167
Preface

Machine learning is increasingly recognized for playing a crucial role in

predicting business performances for frms across the world. On the one
hand, data collection and data wrangling are the important steps in training
machine learning models. On the other hand, business analytics includes
various tools for visualizing and steering data to deliver insights in manage-
ment decisions. An expert business analyst may work with different data
sets, choose appropriate machine learning models, and deliver valuable
insights to improve business performances.
A report published by Market Research Future predicts that the global
machine learning market will grow at over 44% compound annual growth
rate till 2024 to reach $20.83 billion, up from $1.58 billion in 2017. Needless
to say, the adoption of machine learning in business enterprises would cre-
ate enormous opportunities for students and professionals with the required
skill set. This book aims to help machine learning enthusiasts better prepare
for the opportunity.
Overall, this book captures applications of machine learning in different
contexts, such as agriculture, B2B marketing, banking, customer experience,
digital marketing, healthcare, hospitality, insurance, and sustainable develop-
ment. The book is comprised of 13 chapters.
We take this opportunity to express our sincere gratitude to the authors
of the 13 chapters in this book. The authors include academicians from
renowned institutions and industry experts with relevant experience. We
believe that this book will be of signifcant value to different interest groups
that wish to explore the applications of machine learning. Finally, we whole-
heartedly thank our editorial team for extending persistent support to pub-
lish this book.

vii
Editors’ Biographies

Hemachandran K.
Dr. Hemachandran K. is a Professor of Artifcial
Intelligence at the School of Business, Woxsen
University, Hyderabad, Telangana, India. He is
a passionate teacher with 14 years of teaching
experience and 5 years of research experience. He is a
strong educational professional with a scientifc bent of
mind, highly skilled in AI and ML. After receiving his
PhD in embedded systems from Dr. MGR Educational
& Research Institute, India, he conducted interdisciplinary research in AI.
He is an open-ended positive person who has a stupendous peer-reviewed
publication record with more than 20 journals and international conference
publications. He served as an effective resource person at various national
and international scientifc conferences. He has rich research experience in
mentoring UG and PG student projects. He owns two patents and has life
membership in esteemed professional organizations. He was a pioneer to
establish the Single Board Computer Lab at Ashoka Institutions, Hyderabad,
India. His self-paced learning schedule and quest to upgrade and update
learning skills resulted in receipt of 15 online certifcate degrees conferred
by COURSERA and other online platforms. He is also an editorial board
member for numerous reputed SCOPUS/SCI journals.

ix
x ◾ Machine Learning for Business Analytics

Sayantan Khanra
Dr. Sayantan Khanra serves as Assistant Professor
at the School of Business Management, Narsee
Monjee Institute of Management Studies, Mumbai,
India. He holds a PhD in strategic management from
the Indian Institute of Management Rohtak. His
research interests relate to the domains of the digital
economy, management of technology, and sustainable
development. Previously, he worked at the Turku School of Economics,
Finland; the National Taiwan University of Science and Technology, Taipei;
and Woxsen University, India. His research studies are published in Business
Strategy and the Environment, Enterprise Information Systems and the Journal
of Business Research, among other quality academic journals.

Raul V. Rodriguez
Dr. Raul V. Rodriguez is Pro-Vice-Chancellor at
Woxsen University and Dean of the School of
Business at Woxsen University. He holds a PhD in
artifcial intelligence and robotics process automation
applications in Human Resources. He is former
Co-CEO of Irians Research Institute, a research
facility specialized in neuromarketing, AI, ML, market
research, behavioral science, social research, and
behavioral engineering. His areas of expertise and interest are machine
learning, deep learning, natural language processing, computer vision,
robotic process automation, multi-agent systems, knowledge engineering,
and quantum artifcial intelligence. He is profcient in Prolog, Java, C++,
Python, R/RStudio, Julia, Swift, Scala, MySQL, and Spark, among others. He
is a registered expert in artifcial intelligence, intelligent systems, and multi-
agent systems at the European Commission, a nominee for the Forbes 30
under 30 Europe 2020 list, and an awardee in the Europe India 40 under 40
leaders. Additionally, he is a member of the GRLI Deans and Directors cohort.
He has co-authored two reference books: New Age Leadership: A Critical
Insight and Retail Store’e and has more than 70 publications to his credit. He
is a weekly contributing writer to various magazines in the feld of analytics
and emerging technologies. He is also a journal reviewer and associate editor
in various publications such as IEEE.
 Editor’s Biographies ◾ xi

Juan R. Jaramillo
Juan R. Jaramillo is Associate Professor and Academic
Director of the Master in Business Analytics in the
Robert B. Willumstad School of Business at Adelphi
University. He has a PhD and an MS in industrial
engineering from West Virginia University, and he
earned a BS in civil engineering and a BS in geological
engineering from Escuela de Ingeniería de Antioquia
in Medellín Colombia. He is also a Lean Six Sigma
Master Black Belt. At Adelphi, Juan led the design of the Master of Science
in Business Analytics (MSBA). He previously worked at Farmingdale State
College in New York and Albany State University in Georgia. In these
institutions, he designed an undergraduate program in business analytics and
two supply chain programs. Early in his career Juan worked for Colceramica
S. A., one of the largest ceramic producers in the Americas, in the areas of
R&D and manufacturing. He has been involved in the Informs Innovative
Applications in Analytics Award (IAAA) since its inception in 2012 and has
served as a judge, chair (2017–2019), and cochair (2020–2022). He received
the inaugural Michael Gorman Award for his contribution to the Analytics
Society of INFORMS in 2020. His areas of expertise are analytics, artifcial
intelligence, and operations. He has multiple publications in these areas, and
he has been the keynote speaker and panelist at conferences in the United
States, Asia, and Latin America.
Contributors

M. Anand S. Deepajothi
Department of ECE Department of CSE
Dr. MGR Educational & Research Nagarjuna College of Engineering
Institute University and Technology
Chennai, India Bangalore, India

Meghraj Arli Viplav Dhandhukia

School of Business Business Analyst
Woxsen University Simply Fresh Pvt. Ltd.
Hyderabad, India Hyderabad, India

Darsana S. Babu Bijay Kumar G.

Department of ECE School of Business
Baselios Mathews II College of Woxsen University
Engineering Hyderabad, India
Sasthamcotta, Kollam, Kerala, India
Sanjeev Ganguly
Juvairia Begum School of Business
Department of CSE Woxsen University
LORDS Institute of Engineering & Hyderabad, India
Technology
Hyderabad, India Preetha Mary George
Dr. MGR Educational Research
Debdutta Choudhury Institute University
School of Business Chennai, India
Woxsen University
Hyderabad, India

xiii
xiv ◾ Machine Learning for Business Analytics

C. Guzmán-Velásquez J. G. Lalinde-Pulido
Universidad EAFIT Universidad EAFIT
Medellin, Colombia Medellin, Colombia

L. K. Indumathi Megha Mankal

Department of CSE School of Business
LORDS Institute of Engineering & Woxsen University
Technology Hyderabad, India
Hyderabad, India
Praveen Kumar Munari
Syed Hasan Jafar School of Business
School of Business Woxsen University
Woxsen University Hyderabad, India
Hyderabad, India
Gaurav Nagpal
Muneza Kagzi Birla Institute of Technology and
Assistant Professor, Strategy & Science
Sustainability Pilani, India
T. A. Pai Management Institute
Manipal, Karnataka, India B. Justus Rabi
Principal
A. Kannan Christian College of Engineering and
Dean, Energy Conservation and Technology
Electrical Management Tamilnadu, India
Dr. MGR Educational & Research
Institute University H. Raghupathi
Chennai, India Department of ECE
Visvesvaraya College of Engineering
Chinnapani Kiran Kumar and Technology
School of Business Hyderabad, India
Woxsen University
Hyderabad, India Abdul Rais
Department of CSE
Korupalli V. Rajesh Kumar LORDS Institute of Engineering &
School of Business Technology
Woxsen University Hyderabad, India
Hyderabad, India
 Contributors ◾ xv

Princy Sera Rajan Laxmi Shaw

Department of ECE Assistant Professor
Baselios Mathews II College of Department of ECE
Engineering Chaitanya Bharati Institute of
Sasthamcotta, Kollam, Kerala, India Technology
Hyderabad, India
Namita Ruparel
Department of HR Dr. M. A. Sikandar
IBS Hyderabad School of Commerce and Business
IFHE University Management
Hyderabad, India Maulana Azad National Urdu
University
Sameena M. H. Hyderabad, India
Department of AEI
Baselios Mathews II College of Kshitiz Sinha
Engineering Analyst
Sasthamcotta, Kollam, Kerala, India IQVIA
BAaaS Consultant, IQVIA,
Thakur Santosh Bengaluru, India
School of Technology
Woxsen University Harshitha Sirineni
Hyderabad, India School of Business
Woxsen University
Meganathan Kumar Satheesh Hyderabad, India
Panasonic Life Solutions India Pvt Ltd
Chennai, India Venkat Reddy Yasa
School of Business
Himanshu Seth Woxsen University
Indian Institute of Management Hyderabad, India
Rohtak, India

Cynthia Jabbour Sfeir

Offce of Finance Notre Dame
University–Louaize NDU
Zouk Mosbeh, Lebanon
Chapter 1

Introduction to Machine
Learning for Data Analytics
L. K. Indumathi, Abdul Rais, and Juvairia Begum

Contents
1.1 Introduction ................................................................................................2
1.2 Basics of Data.............................................................................................3
1.2.1 What Is Data? .................................................................................3
1.2.2 What Is Data Analysis? ...................................................................4
1.2.3 Why Is Data Analysis Required? ....................................................5
1.2.3.1 Types of Data Analysis ....................................................5
1.2.3.1.1 Descriptive Analysis .......................................5
1.2.3.1.2 Diagnostic Analysis ........................................5
1.2.3.1.3 Predictive Analysis .........................................6
1.2.3.1.4 Prescriptive Analysis ......................................6
1.2.4 Base of Data Mining ......................................................................6
1.2.4.1 Data Processing................................................................6
1.2.4.1.1 Types of Data Processing...............................7
1.2.4.1.2 Data Pre-processing .......................................8
1.2.4.2 Data Cleaning ..................................................................8
1.2.4.2.1 What Is Data Cleaning? ..................................9
1.2.4.2.2 Comparison of Data Transformation and
Data Cleaning .................................................9
1.2.4.2.3 Method to Clean Data ....................................9
1.2.4.2.4 Essential Elements for Quality Data ............10
1.2.4.2.5 Uses of Data Cleaning .................................10

DOI: 10.4324/9781003206316-1 1
2 ◾ Machine Learning for Business Analytics

1.2.4.3
Data Exploratory ............................................................10
1.2.4.3.1 Requirement of Exploratory Data Analysis .... 11
1.2.4.4 Data Visualization ..........................................................11
1.2.4.4.1 Essential of Data Visualization ....................11
1.2.5 Introduction to Machine Learning ...............................................12
1.2.5.1 Necessity of Machine Learning .....................................12
1.2.5.2 Applications of Machine Learning ................................13
1.2.5.2.1 Machine Learning Usage in Industries ........13
1.2.5.2 Relationship between Machine Learning
and Data Analysis ..........................................................14
1.2.5.3 Necessity of Probability for Machine Learning .............14
1.2.5.4 Three Types of Machine Learning Algorithm............... 15
1.2.5.4.1 Supervised Learning .................................... 15
1.2.5.4.2 Unsupervised Learning ................................16
1.2.5.4.3 Reinforcement Learning ...............................16
1.3 Conclusion ................................................................................................16

1.1 Introduction
In the 1960s, computer science became an academic discipline. Many basic
computer science subjects like computer architecture, operating systems, and
computer networks, underpins them and were all highlighted. The study of
algorithms was added as an important component of theory in the 1970s.
The goal was to make computers more useful. Today, a fundamental shift
is taking place, with the emphasis shifting to a wide range of applications.
There are a variety of explanations behind this shift. Computing and com-
munications have become increasingly intertwined. In the natural sciences,
commerce, and other sectors, the increased ability to monitor, acquire, and
store data necessitates a shift in our understanding of data and how to
handle it in the modern era. The rise of the internet and social media as
important components of daily life brings theoretical opportunities as well
as challenges.
Whereas traditional areas of computer science will continue to be impor-
tant, future researchers will be more concerned with using computers to
understand and extract usable data from multiple data generated by appli-
cations, rather than just how to make computers helpful for specifc well-
defned problems.
 Machine Learning for Data Analytics ◾ 3

Digital data is frequently presented with a large number of components

in domains as varied as cognition, retrieval, and machine learning (ML).
The topic model is more than just a way to keep track of multiple felds in a
record.
One of the most surprising developments in computer science in the last
two decades is that some domain-independent methods have proven to be
quite effective in solving problems from a variety of felds. A good example
is ML.
Analysts can derive insights from data through statistical analysis. Data is
analyzed using both statistics and ML approaches. Big Data is employed in
the development of statistical models that show data trends. These models
can then be used to create predictions and inform decision-making using
new data. This procedure requires statistical programming languages like R
or Python (with pandas). Advanced analysis is also possible because to open
source libraries and packages like Tensor Flow.
The process of analyzing, cleaning, manipulating, and modeling data with
the objective of identifying usable information, informing conclusions, and
assisting decision-making is known as data analysis. Data analysis has several
dimensions and approaches, including a wide range of techniques under vari-
ous titles and being applied in a variety of business, science, and social sci-
ence sectors. Data analysis is important in today’s business environment since
it helps businesses make more scientifc decisions and run more effciently.
Chapter 1 gives an overall view of data analysis, data processing, data
cleaning, data visualizing, requirement of ML, probability of ML, and basic
algorithm of ML.

1.2 Basics of Data

1.2.1 What Is Data?
Data in computing knowledge has been converted into a format that is easy
to transfer or process. Data is information translated into binary digital form,
as it relates to today’s computers and transmission media. It is allowed to use
data as either a solitary or plural subject. The term “raw data” refers to data
in its most basic digital version.
The terms “data processing” and “electronic data processing,” although
for a time were used interchangeably to refer to the entire range of what
4 ◾ Machine Learning for Business Analytics

was then recognized as digital technologies, indicated that data analysis is

important in computer-supported collaborative. In the history of corporate
computing, specialization has occurred, and a distinct data profession has
emerged in line only with the development of enterprise data handling.
Computers represent data like video, images, audio, and text as binary
values, which are made up of simply two numbers: 1 and 0. A bit is the
simplest data unit, with a single value. Eight binary digits make up a byte.
Megabytes and gigabytes are capacity and storage units.
As the amount of data collected and stored expands, so do the units of
data measurement. For example, the phrase “brontobyte” refers to data stor-
age equal to 10 to the 27th power of bytes.
Information will be stored in fle types, identical to just how mainframe
systems employ ISAM and VSAM. Some other data format for data storage,
transmission, and analysis is comma-separated values. Further specialization
occurred as a database, a database management system, and then relational
database technologies appeared to organize information.
Over the last decade, the rise of the internet and smartphones has
resulted in a boom in digital data production. Text, audio, and video data,
as well as register and online activity records, are now included in the data.
Unstructured data makes up a large portion of this.
Data of the petabyte or larger range has been referred to as “Big Data.”
The 3Vs—volume, variety, and velocity—describe large data in a simplifed
way. Big Data–driven business models have evolved as web-based e-com-
merce has increased in popularity, recognizing data as an important com-
modity for its own sake.
Outside of its use in data-processing computing applications, data has a
value. In electrical component connectivity and network communication, the
term “data” is often distinguished from “control information,” “control bits,”
and related expressions to describe the core substance of a transmission unit.
Furthermore, in science, the term “data” refers to a collection of facts. This can
be seen in fnance, marketing, demographics, and healthcare, to name a few.

1.2.2 What Is Data Analysis?

Working with data to extract relevant information that can subsequently be
utilized to make informed decisions is known as data analysis.
This concept is the foundation of data analysis. We are better able to
make decisions when we can extract meaning from facts. And we live in an
era where we have access to more data than ever before.
 Machine Learning for Data Analytics ◾ 5

1.2.3 Why Is Data Analysis Required?

In the business world, data analysis is critical for understanding challenges
and exploring data in meaningful ways. Data is nothing more than num-
bers and facts. Data analysis is the process of organizing, interpreting, struc-
turing, and presenting data into valuable information.
Everyone realizes that the goal of data analysis is to help you make data-
driven business choices, otherwise why would you let it take so long that
the results are obsolete by the time you get them? Web data integration auto-
mates all processes of web data analysis, allowing you to gain insights from
data as soon as it is collected. You can use real-time data insights instead of
obsolete insights as a foundation for your company decisions.

1.2.3.1 Types of Data Analysis

Data can be utilized in a variety of ways to answer questions and assist
choices. These types of analyses can be categorized into four groups that are
regularly employed in the feld. We’ll go through each of these data analysis
techniques, as well as an example of how they could be used in the real
world.

1.2.3.1.1 Descriptive Analysis

Big Data and data science have become popular terms in recent years. They
tend to be well-researched, which necessitates careful processing and analy-
sis of the data. Descriptive analysis is one of the approaches used to ana-
lyze this data. What transpired is revealed through descriptive analysis. This
sort of analysis uses statistics to describe or summarize quantitative data.
Statistical analysis, for example, might reveal the distribution of sales among
a group of students as well as the average marks per student. Explanation of
“What Happened” is referred as descriptive analysis.

1.2.3.1.2 Diagnostic Analysis

The “what” is determined by descriptive analysis, whereas the “why” is
determined by diagnostic analysis. Now, let us imagine a descriptive analysis
reveals that a hospital is experiencing an extraordinary infux of patients. If
you go deeper into the data, you might fnd that many of these individuals
have the same virus symptoms. This diagnostic study can help you fgure
out if the infow of patients was caused by an infectious pathogen—the
“why.” Explanation of “Why It Happened” is referred as diagnostic analysis.
6 ◾ Machine Learning for Business Analytics

1.2.3.1.3 Predictive Analysis

As of now, we’ve examined the methods of analysis that look at the past
and draw conclusions. Predictive analytics makes predictions about the
future based on data. You might notice that a particular product has had its
strongest sales during the months of September and October each year, lead-
ing you to predict a similar high point during the future year using predic-
tive analysis. Explanation of “Future Status (What May Happen?)” is referred
as predictive analysis.

1.2.3.1.4 Prescriptive Analysis

Prescriptive analysis combines the fndings of the preceding three forms of anal-
ysis to make ideas for how a corporation should proceed. Applying our analogy,
this form of analysis might recommend a business strategy to capitalize on the
accomplishment of the high-sale months while also identifying fresh growth
prospects during the weaker months. Explanation of “What Is the Reaction” is
referred as diagnostic analysis. It will support the decision-making mechanism.

1.2.4 Base of Data Mining

Data mining is a systematic analytical process for extracting information
from huge amounts of original data. It is a discipline of computer engineer-
ing and analytics that uses sophisticated approaches to fnd hidden patterns
in massive amounts of data. The automation of this type of analysis, which
uses ML and database technologies to speed up the analytical process and
locate knowledge which is more relevant and engaging, is one of its dis-
tinguishing features. Facts mining, contrary to its term, is not usually about
extracting pure data from a mountain of data but rather about detecting
relevant regularities that arise again from data set.
The term “data mining” refers to the process of extracting information
from large amounts of data. To put it another way, data mining is the process
of extracting knowledge from data. The information or knowledge obtained
in this manner can be applied to any of the following purposes like market
analysis, web surfng, control of production, and science exploration.

1.2.4.1 Data Processing

Although most people are familiar with the term “word processing,” comput-
ers were designed for “data processing”—the organizing and manipulation of
 Machine Learning for Data Analytics ◾ 7

enormous volumes of quantitative data, or “number crunching” in computer

jargon. Calculation of satellite orbits, weather forecasting, statistical analysis,
and, in a more practical sense, business applications such as accounting,
payroll, and billing are all instances of data processing.

1.2.4.1.1 Types of Data Processing

Depending on the purpose of the data, many types of data processing
procedures are available. The fve major methods of data processing are
discussed in this book.

(1) Large-Scale Data Processing or Commercial Data Processing Large-

scale data processing is a method of using relational databases in a com-
mercial setting, which involves batch processing. It entails feeding the
system a vast amount of data and producing a signifcant volume of output
with fewer processing operations. It essentially integrates commerce and
computers in order to make it helpful for a company. Because the data
processed by this system is usually standardized, there is a signifcantly
smaller risk of errors.
Many manual tasks are automated using computers to make them easier
and more error-free. In the business world, computers are employed to turn
raw data into information that is benefcial to the company. Accounting soft-
ware is a good example of a data processing application. The feld of infor-
mation systems (IS) studies subjects like organizational computer systems.

(2) Research or Scientifc Data Processing Scientifc or research data pro-

cessing, unlike commercial data processing, makes extensive use of com-
puting operations while requiring fewer inputs and outputs. Arithmetic and
comparison operations are among the computing operations. Any chance
of errors is unacceptable in this form of processing since it would lead to
erroneous decisions. As a result, the process of validating, categorizing, and
standardizing the data is carried out with great care, and a variety of sci-
entifc procedures are employed to ensure that no incorrect associations or
conclusions are made.

(3) Group Processing Group processing is a sort of data processing that

involves processing multiple cases at the same time. It is most commonly uti-
lized when the data is homogeneous and in big amounts, and it is collected
and analyzed in batches. Ongoing, sequential, or chronological processing
8 ◾ Machine Learning for Business Analytics

of an activity is referred to as batch processing. Simultaneous group process-

ing occurs when all of the cases are processed at the same time by the same
resource. Sequential batch processing occurs when various cases are pro-
cessed by the same resource either simultaneously or sequentially.

(4) Electronic Procession The answer to each point and click is computed
long before the user even opens the application in most electronic analytical
processing systems. In truth, many online processing systems conduct the
computation ineffciently, but because the processing is done ahead of time,
the end-user is unaware of the issue. When data must be processed on a
continuous basis and is fed into the system automatically, this sort of pro-
cessing is utilized.

(5) Real-Time Processing The existing data management system often

limits the capacity of processing data on an as-needed basis because it is
always based on periodic batch updates, resulting in a time lag of many
hours between an event occurring and it being recorded or updated. This
necessitated the development of a system that could record, update, and
process data on an as-needed basis, that is, in real-time, decreasing the time
delay among event and execution to nearly zero. Huge amounts of data are
being pumped into the systems of businesses; therefore, storing and analyz-
ing it in real time would be a game changer.

1.2.4.1.2 Data Pre-processing

Pre-processing data is a data mining technique for transforming raw data
into a usable and productive format. Data pre-processing entails converting
raw data into well-formed data sets in order to use data mining methods.
Raw data is frequently incomplete and formatted inconsistently. The success
of every project involving data analytics is directly proportional to the qual-
ity of data preparation.

1.2.4.2 Data Cleaning

Data cleaning, also known as data scrubbing, is one of the most important
features required to create an entity that contains the art of excellent deci-
sion-making. There’s no denying that an analysis can only be good if the
data it is based on is of high quality. The act of producing data for specif-
cation by removing or reshaping data that is missing, wrong, inappropriate,
unsuitable, or redundant is known as data cleaning.
 Machine Learning for Data Analytics ◾ 9

1.2.4.2.1 What Is Data Cleaning?

The practice of updating or removing incorrect, duplicate, corrupted, or
incomplete data from a database is known as data cleaning. If data is
wrong, even if it appears to be correct, algorithms and outputs are unreli-
able. The data cleaning process isn’t only about eliminating data to make
room for new data; it’s also about fguring out how to maximize the validity
of a data collection without erasing it.
Data cleaning entails correcting grammatical and spelling errors, repair-
ing faults like omitting codes and empty felds, fnding duplicate data
points, and standardizing data sets, among other things. It is seen as a basic
aspect of the data science fundamentals and plays a signifcant role in creat-
ing trustworthy results and in the analysis procedure. Data cleaning services
are designed to create uniform and standardized data sets that give data
analytical devices and predictive analytics good accessibility to and percep-
tion of precise fgures for every challenge.

1.2.4.2.2 Comparison of Data Transformation and Data Cleaning

Data warehouses aid in data analysis, report creation, data visualization,
and business decision-making. In data warehousing, two strategies are used:
data transformation and data cleaning. Data cleaning refers to the removal
of incoherent data from a database in order to improve data homogeneity,
whereas data transformation refers to the translation of data from one struc-
ture to another in order to facilitate processing.

1.2.4.2.3 Method to Clean Data

A data cleaning method will change most parts of an entity’s general data
cleaning program, but it is only one part of a long-term data cleaning solu-
tion. The following are the steps for data cleaning

Step 1: The frst step is to determine which types of data felds are
required for the project.
Step 2: The information in the data felds that have been short-listed is
gathered, categorized, and organized.
Step 3: Duplicate fgures are identifed and removed, and mistakes are
corrected.
Step 4: Tools for data purifcation is to avoid information gaps, look for
and fll in the missing values in the data collection.
Step 5: The data operation should be standardized based on repeated
tests and methodologies that have shown to create high-quality data,
10 ◾ Machine Learning for Business Analytics

allowing for easy replication and consistency later on. It is a need for
the person responsible for the maintaining process that the procedure
and frequency of data cleaning be standardized taking into account the
most often used data.
Step 6: Every week or month, a specifc amount of time must be set aside
to closely examine the faws, approaches that are performing effectively,
areas for improvement, and errors and glitches that are occurring.

1.2.4.2.4 Essential Elements for Quality Data

Establishing the standard of information necessitates an examination of its
qualities, followed by a ranking of those traits in terms of their value and
applicability in the company. The following are the qualities that must be
present in high-quality data:

Authenticity: The amount to which the data conforms to stated business

restrictions and regulations.
Correctness: The information must be able to depict the real and best
values.
Exactness: The amount to which you are familiar with all of the required
info.

Data synchronizes inside the same database or across distinct databases and
is referred to as data synchronization. Homogeneity refers to how closely the
data adheres to the same measurement units.

1.2.4.2.5 Uses of Data Cleaning

Obtaining clean and quality data will almost certainly boost overall produc-
tivity and provide high-quality data for rapid and accurate decision-making.
Because many sources of data are involved, errors are eliminated to ensure
seamless operation. Clients will be happier and more satisfed if there are
few to no errors, and employees will be less stressed. It’s easier to trouble-
shoot inaccurate data for future applications when you keep track of prob-
lems and have a higher standard of reporting and pinpointing the source of
issues.

1.2.4.3 Data Exploratory

Exploratory data analysis (EDA) is used by data scientists to assess, study,
and describe data sets’ primary attributes, often employing data visualization
 Machine Learning for Data Analytics ◾ 11

tools. It makes it easier for data scientists to uncover patterns, test hypoth-
eses, and check assumptions by assisting them in determining how to best
manipulate data sources to achieve the answers they require.
EDA is commonly utilized and see what stats can reveal outside of
conventional planning or hypothesis testing procedures, as well as to
learn more about data set factors and their interactions. It could also help
you fgure out if the analysis tools you’re considering for data analysis are
appropriate.

1.2.4.3.1 Requirement of Exploratory Data Analysis

The main purpose of EDA is to help with data analysis before jumping to
any conclusions. It can help in the spotting of obvious mistakes, providing a
better understanding of data trends, detecting extremes or unforeseen issues,
and discovering fascinating correlations between variables.
Data scientists can employ an investigative study to determine whether
the results produced are reliable and acceptable for any specifc business
goals and outcomes. EDA also supports stakeholders by ensuring that the
right questions are being asked. It can address questions about standard
deviations, predictor variables, and error bars.

1.2.4.4 Data Visualization

The graphical depiction of information and data is known as data visual-
ization. Data visualization tools make it easy to examine and comprehend
trends, outliers, and patterns in data by employing visual elements like
charts, graphs, and maps. Data visualization tools and technologies are criti-
cal in the Big Data environment for analyzing enormous volumes of data
and making information choices.

1.2.4.4.1 Essential of Data Visualization

It’s diffcult to imagine a career that doesn’t result in better data under-
standing. Analyzing data is benefcial to every STEM profession, as well as
felds such as administration, business, advertising, olden times, consumer
products, skilled trades, education, and sports. Even though we’ll always
have rhapsodies about data visualization, there are clear practices. Because
visualization is so common, it’s still one of the most valuable career tal-
ents to learn. The more you can graphically communicate your arguments,
whether it’s in a display or a fash deck, the more effectively you could use
that data.
12 ◾ Machine Learning for Business Analytics

1.2.5 Introduction to Machine Learning

ML is a feld that straddles computer science, engineering, and statistics,
with applications in a variety of felds. It can be used in a variety of sectors,
from politics to geosciences, as you’ll discover later. It’s a versatile tool that
can be used to solve a variety of issues. ML techniques can be used in any
industry that requires data interpretation and action. Statistics are used in
ML. To the majority of individuals, statistics is a technical topic employed by
businesses to exaggerate the value of a product. We have adequate com-
puter resources to model the problem properly. We need statistics to solve
these issues. Human motivation, for example, is a challenge that is now
impossible to model. Engineering is the application of science to the solu-
tion of a problem. We’re used to solving deterministic problems in engi-
neering, wherein every answer always solves the problem. If we’re writing
software to control a vending machine, it needs to work all of the time,
regardless of how much money is put in or how many buttons are touched.
There are numerous problems for which there is no predictable solution.
To put it another way, we don’t know enough about the problem or don’t
have enough computational capacity to simulate it properly. We need statis-
tics to solve these issues. Concept of motivation, for example, is a challenge
that is now impossible to model.

1.2.5.1 Necessity of Machine Learning

The progress of ML has been fueled by the virtually infnite amount of
data available, affordable data storage, and the development of less expen-
sive and more powerful computing. Many sectors are now building more
powerful models that can analyze more and more complicated data while
delivering faster and higher accuracy on massive sizes. Organizations can
use ML tools to detect proftable possibilities and potential dangers more
quickly.
ML’s practical applications produce business outcomes that can have a
signifcant impact on a frm’s proftability. Innovative approaches in the dis-
cipline are continually advancing, allowing for practically unlimited appli-
cations of ML. Companies that rely on large amounts of data and require a
system to evaluate it quickly and accurately have adopted ML as the most
effective technique to create models, strategize, and plan.
 Machine Learning for Data Analytics ◾ 13

1.2.5.2 Applications of Machine Learning

1.2.5.2.1 Machine Learning Usage in Industries
a. Healthcare: Portable Iot sensors that track all from heart rates and
steps taken to oxygen and serum glucose and even sleep habits have
created a large amount of data that allows doctors to examine their
patients’ health in real time. A new ML algorithm detects malignant
tumors on mammograms, identifes skin cancer, and analyzes retinal
images to identify diabetic retinopathy, to name a few examples.
b. Government: Government offcials can utilize data to foresee possible
future scenarios and adjust to fast-changing events using ML systems.
ML can aid in fltering citizen policy data and building of automatic
alerts for the executives about the rules and regulation that they need to
remember.
c. Cybersecurity and cyber intelligence assist in counter-terrorism activi-
ties, the optimization of operational capability, warehouse management,
and forecasting, as well as the reduction of failure rates. The healthcare
industry has ten more uses for ML, according to a recent article.
d. Sales and Marketing: The ML algorithm learns the data patterns from
the historical sales data and uses them to predict the future sales and
even gives insights to increase the sales.

Many frms have effectively adopted artifcial intelligence (AI) and ML to

raise and enhance customer satisfaction by over 10%, demonstrating that
ML is transforming the marketing feld. According to Forbes, “57 percent of
enterprise leaders feel that increasing customer experiences and support will
be the most critical growth beneft of AI and ML.”
E-commerce and social networking sites employ ML to evaluate your pur-
chase and search history and offer recommendations for other goods to buy
based on your previous purchases. Most researchers estimate that AI and
ML will drive the future of retail as systems improve their ability to capture,
analyze, and use data to tailor customers’ customer experience and generate
personalized, appropriate marketing strategies.
Shipping: Proftability in this industry depends on effciency and preci-
sion, as well as the capacity to forecast and manage possible diffculties.
Data analysis and modeling capabilities of ML are ideal for businesses in the
delivery, public transportation, and freight transportation industries. ML is a
14 ◾ Machine Learning for Business Analytics

vital component of supply chain management since it employs algorithms to

uncover elements that favorably and negatively affect the success of a supply
chain. During shipping, ML enables scheduling algorithms to optimize pro-
vider selection, rating, routing, and quality control operations, saving money
and increasing effciency. Because of its ability to evaluate thousands of data
points at once and apply algorithms faster than a human, ML can address
issues which are yet to be defned.

1.2.5.2 Relationship between Machine Learning

and Data Analysis
The opportunity to comprehend and identify prospects and customers rises
in tandem with the growth of consumer data.

◾ Businesses must defne their strategy strategically to capitalize on this

data. After all, simply having the information isn’t enough to:
– interpret and comprehend the story being told,
– assess whether information would be most useful to which client,
and
– inspire in employees a culture of data discovery, especially if acting
on hunches is a habit.
In this way, analytics software that naturally fosters data-driven deci-
sion-making gives you a leg up on the competition.

The introduction of AI analytics has altered the conversation’s foundation.

Analytics tools are not only enablers of data gathering but also able to com-
plete the real labor that was once unique to people, thanks to AI’s automa-
tion and augmentation capabilities.
Business leaders recognize the importance of data customized for each
function, as well as the role analytics tools play in the entire workforce of
using that data.

1.2.5.3 Necessity of Probability for Machine Learning

It’s possible to infer that probability is required to effectively complete an ML
predictive modeling project. The technique of developing prediction models
from contradictory data is known as ML. Uncertainty typically works with
incomplete or inaccurate facts.
 Machine Learning for Data Analytics ◾ 15

Though inconsistency is critical to ML, it is now one of the aspects that

beginners, notably those with a programming experience, fnd the most
challenging.
There are three main sources of ambiguity in ML: noisy data, poor pen-
etration of the issue region, and inappropriate algorithms.
Nonetheless, we may use probability methods to handle ambiguity. To
control the ambiguity we see in each project as ML practitioners, we need to
know probability.
Machine learning is built on the foundation of probability.

◾ Model-based approaches must forecast the likelihood of belonging to a

given class.
◾ Probability is used to build algorithms (e.g. Naive Bayes).
◾ Probability will be used by learning algorithms to make decisions (e.g.
information gain).
◾ Probability is the foundation of subfelds of study (e.g. Bayesian
networks).
◾ Probability frameworks are used to train algorithms (e.g. maximum
likelihood).
◾ Probabilistic loss functions are used to ft models (e.g. log loss and cross
entropy).
◾ Probability is used to confgure model hyperparameters (e.g. Bayesian
optimization).
◾ Design skill is assessed using probabilistic methods.

1.2.5.4 Three Types of Machine Learning Algorithm

Commonly used ML algorithms are given specifc attention. The algorithms
studied include linear regression, logistic regression, Naive Bayes, kNN,
random forest, and others. The basic three types of ML algorithms are as
follows.

1.2.5.4.1 Supervised Learning

This algorithm is made up of a target/result variable (or dependent variable)
that must be estimated from a set of predictor variables (independent vari-
ables). We create a function that maps inputs to desired outputs using this
set of variables. The model is trained until it accomplishes the appropriate
16 ◾ Machine Learning for Business Analytics

level of accuracy on the data set. Regression, decision tree, random forest,
KNN, logistic regression, and others are examples of supervised learning.

1.2.5.4.2 Unsupervised Learning

This algorithm is made up of a target/result variable (or dependent variable)
that must be estimated from a set of predictor variables (independent vari-
ables). We create a function that maps inputs to desired outputs using this
set of variables. The model is trained until it accomplishes the appropriate
level of accuracy on the data set. Regression, decision tree, random forest,
KNN, logistic regression, and others are examples of supervised learning.

1.2.5.4.3 Reinforcement Learning

The machine is taught to make certain decisions using this algorithm. It
works like this: the machine is deemed to occur where it must constantly
train itself through trial and error. This computer adapts from its previous
experiences and attempts to gather the most relevant information in order to
make appropriate business decisions.

1.3 Conclusion
ML is a type of data analysis that facilitates the creation of analytical models.
It helps computers to discover underlying truths without becoming specif-
cally instructed wherever to seek through using algorithms that continuously
help in decision-making.
ML now is not the same as ML in the past, thanks to advances in comput-
ing technology. Whereas many ML techniques have indeed been known for
a while, the capacity to apply diffcult mathematical computations to large
amounts of data automatically—again and over, quicker and faster—is a rela-
tively new phenomenon.
Chapter 2

Role of Machine Learning in

Promoting Sustainability
Muneza Kagzi

Contents
2.1 Introduction ..............................................................................................18
2.2 Machine Learning for Planet ....................................................................18
2.2.1 Machine Learning for Planning and Promoting
Sustainable Cities ..........................................................................19
2.2.2 Machine Learning for Waste Management ..................................19
2.2.3 Machine Learning for Clean Energy ............................................19
2.3 Machine Learning for People...................................................................20
2.3.1 Machine Learning for Assisting Green Customers ......................20
2.3.2 Machine Learning to Safeguard the Interest of Shareholders .....21
2.3.3 Machine Learning for e-Government Services for Citizens ........21
2.4 Machine Learning for Proft .....................................................................21
2.4.1 Machine Learning at the Company Level ....................................22
2.4.2 Machine Learning at the Industry Level ......................................23
2.4.3 Machine Learning at the Country Level ......................................23
2.5 Discussion and Conclusion ......................................................................24
References .........................................................................................................25

DOI: 10.4324/9781003206316-2 17
18 ◾ Machine Learning for Business Analytics

2.1 Introduction
There is a universal agreement that proactive measures are needed to save
the planet (Pérez-Ortiz, de La Paz-Marín, Gutiérrez, & Hervás-Martínez,
2014). The importance of sustainability has increased tremendously on the
back of the rising number of natural calamities inficting the planet. The
central idea of sustainability is based on the fact that human consumption
is far exceeding what the planet is able to reproduce. This rising difference
between reproduction and consumption is a cause of worry, shifting focus
to sustainability as a means to bridge the gap. Sustainability has attracted
much attention from researchers, policy-makers, consumers, and investors
across the globe (Zhao, Liu, Zhang, & Fu, 2020). Elkington (1998) conceptu-
alized an approach to achieving sustainability by following the triple bottom
lines that represented the economic, environmental, and social criteria of
corporate performances. To implement sustainable development, researchers
have recommended the implementation of various machine learning (ML)
techniques.
The wider application of ML has gained momentum in recent times.
ML seeks to imitate human learning by using input data, algorithms, and
statistical methods to make predictions or decisions (Fahdi, Elkhechaf, &
Hachimi, 2021). ML techniques identify the pattern in data and produce
an estimated outcome. These techniques help to identify the problem of
resource consumption in the domain of sustainability and subsequently
use data and insights to design optimized resource consumption solutions.
This conceptual chapter conducts an in-depth analysis of existing research
on ML and how the insights can be leveraged to further the concept of
sustainability with respect to the three P’s—namely, people, proft, and
planet.

2.2 Machine Learning for Planet

Environmental sustainability is very crucial for the survival of the planet
(Elkington, 1998). Enterprises need to reduce their carbon footprint in their
business activities and make them more sustainable. An extensive review of
the literature throws up some interesting insights into how enterprises can
reduce their carbon footprint and follow sustainable development practices
in the interest of the planet.
 Promoting Sustainability-Machine Learning ◾ 19

2.2.1 Machine Learning for Planning and

Promoting Sustainable Cities
Urbanization is one of the driving forces of economic growth. However, this
continuous process of urbanization has been at the cost of the environment,
and today, there is more need than ever to follow sustainable development
practices. Li et al. (2019) proposed the use of an ML application, the agent-
based modeling with embedding for geo-simulation, to create sustainable
cities and communities. This methodology uses past location data to make
new location decisions and shape sustainable urban development.
Lan, Zheng, and Li (2021) combined neural networks and deep learning
and in urban planning with a view to incorporate principles of sustainability.

2.2.2 Machine Learning for Waste Management

The management of municipal solid waste (MSW) is a crucial aspect of sus-
tainable development of the urban space. However, MSW sometimes does
not get the attention it deserves mainly due to the lack of adequate waste
management infrastructure. To manage MSW effectively, it is important to
accurately forecast the MSW composition and generation rate. However,
forecasting MSW can be challenging in the context of low-income countries,
where data is either unavailable or unreliable. Research (Ayeleru, Fajimi,
Oboirien, & Olubambi, 2021) recommended two ML techniques—artifcial
neural network (ANN) and supported vector machine (SVM) for resolving
the issue in the context of the City of Johannesburg, South Africa, to forecast
the quantity of MSW generation. Accurate MSW forecasting, therefore, can
go a long way to forecast MSW, thus enabling authorities to plan for ade-
quate waste management and infrastructure.

2.2.3 Machine Learning for Clean Energy

ML can be applied to promote clean energy generation for the beneft of
the planet. Geothermal power plants are a means to generate clean energy.
Geothermal energy refers to the energy that is generated from the rocks
and fuids found beneath the earth’s crust, as far down to the hot molten
rock, magma (Coro & Trumpy, 2020). However, the effectiveness of the
geothermal power plants is dependent on the area where it is located and
also on several environmental factors. ML can be used for renewable energy
20 ◾ Machine Learning for Business Analytics

predictions. The data-driven ML models can be used to create suitability

maps of geothermal sites. Thus, ML can help to identify special geographies
for geothermal power plant installations. The assessment of suitable geog-
raphies for geothermal suitability involves high costs, invasive inspections,
and legal permissions. Research proposes the Maximum Entropy (MaxEnt),
an ecological modeling ML model, to create suitability maps for the set-
ting up of geothermal power plants (Coro & Trumpy, 2020). The MaxEnt
algorithm model can be trained with environmental data and be potentially
correlated with geothermal site suitability and geothermal plant operations.
Such data-driven deep learning models can help to make renewable-energy
predictions, thus saving time, money, effort, and other valuable resources.
ML, in the context of clean energy, can, therefore, equip policy-makers,
geothermal energy industry operators, geologists, and territory citizens with
the relevant data to assess suitable sites and plan for future geothermal
power plants.

2.3 Machine Learning for People

Sustainable development is crucial for the health and well-being of all
people associated with an enterprise, be it the employees or the external
communities and stakeholders, such as suppliers, customers, and sharehold-
ers. Existing research testifes to the following ways in which ML helps to
improve well-being of the people.

2.3.1 Machine Learning for Assisting Green Customers

ML promotes responsible production and consumption. Consumers today
are more environmentally aware and prone to purchase products or ser-
vices that have low environmental impact. More and more brands are now
showcasing their green credentials to build credibility with their custom-
ers. As a case point, the hospitality and tourism sector is increasingly using
ML for sustainable business practices (Khanra, Dhir, Kaur, & Mäntymäki,
2021). For instance, the popular online travel company with user-generated
content feeds consumer feedback data in its ML program to create a frame-
work for recommending green hotels. The ML methods for such predictions
include Self-Organizing Map (SOM), Adaptive Neuro-Fuzzy Inference System
(ANFIS), Higher-Order SVD (HOSVD) technique, and Classifcation And
Regression Tree (CART) (Nilashi et al., 2019).
 Promoting Sustainability-Machine Learning ◾ 21

2.3.2 Machine Learning to Safeguard the Interest of Shareholders

ML can promote shareholders’ interest by providing information on whether or
not the corporates are following ethical practices. Researchers have suggested
the application of random forest (RF) model to fulfll the variable importance
ranking and corporate misconduct prediction (Wang, Asghari, Hsu, Lee, &
Chen, 2020). This model was used in the context of China. It took into con-
sideration observations from construction companies in 2000–2018. The RF
model identifed the key variables of corporate governance. These variables
could be used to forecast corporate misconduct. Such fndings can help pol-
icy-makers to improve decision-making and better regulate malpractices that
adversely impact the environment. Further, they can help regulators in timely
identifying violating companies and implement proactive interventions.

2.3.3 Machine Learning for e-Government Services for Citizens

Prior studies have highlighted that effective use of technology may help gov-
ernment authorities to better serve their citizens (Khanra, Joseph, & Ruparel,
2019). For instance, Khanra and Joseph (2020a) developed a mathematical
model for a smooth delivery of e-Government services. Applications of ML
can help to automatically classify civic queries and ensure their faster disposal
(Kim, Yoo, Park, Lee, & Kim, 2021). Kim et al. (2021) developed a dynamic
topic model to identify the associations among the keywords in civic com-
plaints. The model processes words in text documents and classifes queries
following the Word2vec approach to provide data-driven recommendations
(Kim et al., 2021). However, demographic characteristics and service quality
can impact adoption and dissemination of e-government services (Khanra &
Joseph, 2017), as also English profciency and digital divide (Khanra & Joseph,
2019). Therefore, Kim et al. (2021) tested their ML model with a large data
pool of 1,60,316 civic complaints raised between 2006 and 2017 in Seoul,
South Korea. Furthermore, Khanra and Joseph (2020b) developed an assess-
ment framework for dissemination of effective e-Government services.

2.4 Machine Learning for Proft

Sustainable development refers to the process of generating economic proft
while balancing the environmental and social aspects. To this extent, prior
studies recommend the following solutions.
22 ◾ Machine Learning for Business Analytics

2.4.1 Machine Learning at the Company Level

To generate proft, companies can redesign their existing processes to
make them more cost-effective—as a case point, redesign existing con-
crete machines used in the cement industry. Concrete is largely used
in the construction industry creating environmental challenges such as
energy consumption, depletion of natural resources, and greenhouse gas
emission (Naseri, Jahanbakhsh, Hosseini, & Nejad, 2020). Therefore, there
is a need to redesign eco-friendly versions of these concrete machines
that are cost-effective, have compressive strength, minimize energy and
resource consumption, and reduce greenhouse gas emissions (Naseri
et al., 2020). To design the green concrete machine, the most effcient
objective function is used specifying various sustainability parameters
(Naseri et al., 2020).
Researchers have proposed ML techniques that can predict impor-
tant mechanical characteristics of concrete, for instance, its compres-
sive strength to determine its quality (Naseri et al., 2020). The higher
the compressive strength of the concrete machine, less the amount of
energy consumption and material used, less the amount of greenhouse
gas (embodied CO2) emitted, and ultimately less the budget required for
manufacturing. To forecast compressive strength of eco-friendly concrete,
six ML algorithms were proposed (Naseri et al., 2020). These include the
soccer league competition algorithm, water cycle algorithm (WCA), ANN,
genetic algorithm (GA), regression, and support vector machine (SWM)
(Naseri et al., 2020). Of these, results show that WCA is the most effcient
model for predicting an eco-friendly concrete machine given that it con-
siders mean absolute error and coeffcient of determination. WCA helps
to optimize the use of resources in designing an eco-friendly concrete
machine.
Accordingly, the cost, compressive strength, and environmental impacts,
including embodied carbon dioxide emission, and energy and material
requirements, are considered as sustainability attributes (Naseri et al., 2020).
To integrate their infuence on mixture designing, six objective functions are
applied. These include the fractional-based sustainable objective function, lin-
ear form sustainable objective functions, and distances of sustainability param-
eters among others. As per the name of water cycle algorithm, it is based on
the idea of the process of river fowing into sea from various places. In the
context of designing eco-friendly concrete machines, optimization plays an
important role in utilizing multiple resources and objective functions.
 Promoting Sustainability-Machine Learning ◾ 23

2.4.2 Machine Learning at the Industry Level

Mele and Magazzino (2020) have recommended applications of ML to pro-
mote sustainability in the Chinese steel industry. An ML model named
long short-term memory may be used to balance economic growth and
environmental impact to sustain the steel industry in the long run (Mele &
Magazzino, 2020). The concept of memory featuring the data may be uti-
lized for prediction using gates (Mele & Magazzino, 2020). This model may
examine the relationships among the three important variables, namely pro-
duction in the iron and steel industries, air pollution, and economic develop-
ment (Mele & Magazzino, 2020). The network of long short-term memory is
based on the recursive neural network (Mele & Magazzino, 2020). Therefore,
application of ML may better predict the outcome of steel production on the
economy and environment, helping to balance production and its environ-
ment impact (Mele & Magazzino, 2020). Furthermore, as more and more
manufacturing frms strive to servitize, manufacturing sectors may embrace
greater use of ML to reduce resource consumption (Khanra, Dhir, Parida, &
Kohtamäki, 2021).

2.4.3 Machine Learning at the Country Level

The sugarcane industry’s expansion in Brazil led to deforestation in the
Amazon rainforest, leading to climate change and water and energy inse-
curities. This built the case for the industry to adopt sustainable prac-
tices. In this context, the researchers have used ML techniques to identify
the commitment of an enterprise to sustainability. The Latent Dirichlet
Allocation (LDA) algorithm can be used to extract topics/themes from a
large number of documents, such as websites, sustainability reports, and
fyers (Benites-Lazaro, Giatti, & Giarolla, 2018). LDA can analyze the data
and predict if the company follows sustainable practices. An LDA analy-
sis in the context of the Brazil sugar industry revealed 36 key themes that
demonstrated the initiatives it has taken to promote sustainability. These
include formulating standards and codes of conduct and implementing
various CSR programs. Few of the themes included bioelectricity, agro-
environmental, consumption, biofuels, emission, certifcation, UNICA, stan-
dard, code, climate, labor, committee, report, education, fnancial, fex-fuel,
position, sustainable, health, trade, industry, job, land, voluntary, local, life,
logistic, technology, model, preservation, production, program, riparian,
supplier, tariff, and water.
24 ◾ Machine Learning for Business Analytics

2.5 Discussion and Conclusion

This conceptual chapter examines the role of ML in achieving sustainability.
It aptly demonstrates with examples how ML can contribute to sustainable
development with impacts at three levels, that is, people, proft, and planet.
The study also develops a framework, as presented in Figure 2.1, to posit
that the application of ML tools and techniques will beneft society, the envi-
ronment, and the industries. This framework also outlines the specifc ML
techniques and the context of their application.
For all data-driven predictions related to a sustainable planet, research
recommends the use of agent-based model (Li et al., 2019), deep neural
network, and the random forest model (Lan et al., 2021). Urbanization is
witnessing increased growth on the back of economic growth. Location
planning using ML will enable authorities to develop sustainable urban
solutions. Artifcial neural network and SVM methodology (Ayeleru et al.,
2021) have found use in designing effective waste management solutions.
Maximum entropy (Coro & Trumpy, 2020) has found use in mapping loca-
tions for geo-thermal plants, which can serve as sources for clean energy.
ML methods such as SOM, ANFIS, HOSVD technique, and CART (Nilashi
et al., 2019) are being extensively used in the tourism and hospitality sectors,

Figure 2.1 Application of machine learning for promoting sustainability.

 Promoting Sustainability-Machine Learning ◾ 25

for instance, to recommend green hotel accommodations using user-

generated content as data, as in the case of the e-platform, TripAdvisor.
ML can also promote the well-being of people within and outside an
organization. Fuzzy set theory and ensemble learning technology can help
to manage green suppliers (Wu, Lin, Barnes, & Zhang, 2020). On the other
hand, SOM, ANFIS, HOSVD, and CART have found use in promoting green
customers (Nilashi et al., 2019). Furthermore, random forest methodology
can help to predict unethical practices in a corporate setting, thus helping to
protect shareholders’ rights.
To promote proftability, ML can be implemented at three levels, namely,
product, industry, and country. As a case point, WCA can be used to design a
sustainable cement concrete machine (Naseri et al., 2020). Similarly, LDA can
be implemented at the country level to promote sustainability in the sugar-
cane industry (Benites-Lazaro et al., 2018). Long short-term memory (Mele
& Magazzino, 2020) can be utilized to sustain the steel industry in the long
run. The concept of memory featuring data can be used to forecast the out-
put of the steel industry and measure the impact it has on the environment
and economy. From the previous discussion, it becomes clear that various
ML techniques can be used to combat climate change and promote green
practices in all aspects of our activities. While ML is still at the nascent stage,
given the success it has met with, its use will grow all the more as it evolves.

References
Ayeleru, O. O., Fajimi, L. I., Oboirien, B. O., & Olubambi, P. A. (2021). Forecasting
municipal solid waste quantity using artifcial neural network and supported
vector machine techniques: A case study of Johannesburg, South Africa.
Journal of Cleaner Production, 289, 125671.
Benites-Lazaro, L. L., Giatti, L., & Giarolla, A. (2018). Sustainability and governance
of sugarcane ethanol companies in Brazil: Topic modeling analysis of CSR
reporting. Journal of Cleaner Production, 197, 583–591.
Coro, G., & Trumpy, E. (2020). Predicting geographical suitability of geothermal
power plants. Journal of Cleaner Production, 267, 121874.
Elkington, J. (1998). Partnerships from cannibals with forks: The triple bottom line
of 21st-century business. Environmental Quality Management, 8(1), 37–51.
Fahdi, S., Elkhechaf, M., & Hachimi, H. (2021). Machine learning for cleaner pro-
duction in port of Casablanca. Journal of Cleaner Production, 294, 126269.
Khanra, S., Dhir, A., Kaur, P., & Mäntymäki, M. (2021). Bibliometric analysis and
literature review of ecotourism: Toward sustainable development. Tourism
Management Perspectives, 37, 100777.
26 ◾ Machine Learning for Business Analytics

Khanra, S., Dhir, A., Parida, V., & Kohtamäki, M. (2021). Servitization research: A
review and bibliometric analysis of past achievements and future promises.
Journal of Business Research, 131, 151–166.
Khanra, S., & Joseph, R. P. (2017). Adoption and diffusion of e-government ser-
vices: The impact of demography and service quality. In Proceedings of
the 10th International Conference on Theory and Practice of Electronic
Governance (pp. 602–605). ACM Digital Library, New York, NY, USA.
Khanra, S., & Joseph, R. P. (2019). Adoption of e-governance: The mediating role
of language profciency and digital divide in an emerging market context.
Transforming Government: People, Process and Policy, 13(2), 122–142.
Khanra, S., & Joseph, R. P. (2020a). Orchestration of an e-government network:
Capturing the dynamics of e-government service delivery through theoretical
analysis and mathematical forecasting. In International Working Conference
on Transfer and Diffusion of IT (pp. 219–229). Springer, Cham.
Khanra, S., & Joseph, R. P. (2020b). Development and validation of an assess-
ment framework for e-government services. In International Conference on
Electronic Governance and Open Society: Challenges in Eurasia (pp. 27–41).
Springer, Cham.
Khanra, S., Joseph, R. P., & Ruparel, N. (2019). Dynamism of an e-government
network in delivering public services. In Academy of Management Global
Proceedings (p. 376). Academy of Management,USA.
Kim, B., Yoo, M., Park, K. C., Lee, K. R., & Kim, J. H. (2021). A value of civic
voices for smart city: A big data analysis of civic queries posed by Seoul citi-
zens. Cities, 108, 102941.
Lan, H., Zheng, P., & Li, Z. (2021). Constructing urban sprawl measurement sys-
tem of the Yangtze River economic belt zone for healthier lives and social
changes in sustainable cities. Technological Forecasting and Social Change,
165, 120569.
Li, F., Xie, Z., Clarke, K. C., Li, M., Chen, H., Liang, J., & Chen, Z. (2019). An agent-
based procedure with an embedded agent learning model for residential land
growth simulation: The case study of Nanjing, China. Cities, 88, 155–165.
Mele, M., & Magazzino, C. (2020). A machine learning analysis of the relationship
among iron and steel industries, air pollution, and economic growth in China.
Journal of Cleaner Production, 277, 123293.
Naseri, H., Jahanbakhsh, H., Hosseini, P., & Nejad, F. M. (2020). Designing sus-
tainable concrete mixture by developing a new machine learning technique.
Journal of Cleaner Production, 258, 120578.
Nilashi, M., Ahani, A., Esfahani, M. D., Yadegaridehkordi, E., Samad, S., Ibrahim,
O., Sharef, N. M., & Akbari, E. (2019). Preference learning for eco-friendly
hotels recommendation: A multi-criteria collaborative fltering approach.
Journal of Cleaner Production, 215, 767–783.
 Promoting Sustainability-Machine Learning ◾ 27

Pérez-Ortiz, M., de La Paz-Marín, M., Gutiérrez, P. A., & Hervás-Martínez, C.

(2014). Classifcation of EU countries’ progress towards sustainable develop-
ment based on ordinal regression techniques. Knowledge-Based Systems, 66,
178–189.
Wang, R., Asghari, V., Hsu, S. C., Lee, C. J., & Chen, J. H. (2020). Detecting cor-
porate misconduct through random forest in China’s construction industry.
Journal of Cleaner Production, 268, 122266.
Wu, C., Lin, C., Barnes, D., & Zhang, Y. (2020). Partner selection in sustain-
able supply chains: A fuzzy ensemble learning model. Journal of Cleaner
Production, 275, 123165.
Zhao, S., Liu, Y., Zhang, R., & Fu, B. (2020). China’s population spatialization
based on three machine learning models. Journal of Cleaner Production, 256,
120644.
Chapter 3

Addressing the Utilization of

Popular Regression Models
in Business Applications
Meganathan Kumar Satheesh and Korupalli V. Rajesh Kumar

Contents
3.1 Introduction ..............................................................................................30
3.2 Popular Regression Models ......................................................................31
3.2.1 Simple Linear Regression .............................................................31
3.2.2 Multiple Linear Regression ...........................................................31
3.2.3 Logistic Regression .......................................................................31
3.2.4 Ridge Regression ..........................................................................32
3.2.5 Lasso Regression ..........................................................................32
3.2.6 Random Forest Regression ...........................................................33
3.2.7 Support Vector Regression ...........................................................33
3.2.8 Quantile Regression .....................................................................33
3.2.9 Limitations of Regressions ...........................................................33
3.3 Applications of Regression Models in Four Domains of the Business ..34
3.3.1 Finance .........................................................................................34
3.3.1.1 Credit Score ....................................................................34
3.3.1.2 Risk Analysis ..................................................................35
3.3.1.3 Business Performance....................................................35
3.3.1.4 Revenue Model ..............................................................35
3.3.1.5 Value at Risk ..................................................................35
3.3.1.6 Business Failure .............................................................36

DOI: 10.4324/9781003206316-3 29
30 ◾ Machine Learning for Business Analytics

3.3.2
Marketing ......................................................................................36
3.3.2.1 Direct Marketing Score Model .......................................36
3.3.2.2 Family Business Research ..............................................36
3.3.2.3 Market Consumption and Price Determinants .............37
3.3.2.4 Business Growth ............................................................37
3.3.2.5 Export Market ................................................................37
3.3.3 Human Resource Management ....................................................38
3.3.3.1 Organizational Justice and Employee Commitment .....38
3.3.3.2 Employee Selection ........................................................38
3.3.3.3 Human Capital Development ........................................38
3.3.3.4 Payroll and Compensation ............................................39
3.3.4 Operations ....................................................................................39
3.3.4.1 Demand Forecasting ......................................................39
3.3.4.2 Supply Chain Flexibility.................................................39
3.3.4.3 Green Supply Chain Management ................................40
3.4 Conclusion ................................................................................................40
References .........................................................................................................40

3.1 Introduction
Globalization has made companies change the strategy, both internal and
external, for maintaining a good relationship with business partners and cus-
tomers (Bergeron, Raymond, & Rivard, 2004). This makes the organization
do a lot of prediction and forecasting to achieve the goal. However, a bad
prediction of the business failure will affect the success of the business due
to inappropriate decisions and will reduce proft for the stakeholders due
to a fall in the market value of the frm (Sofe & Hubert, 2004). Prediction
models are also used to understand the connections between the buyers and
sellers, paving the way for comparing the various predictive models to get
higher accuracy (Zuo, Kajikawa, & Mori, 2016).
Statistical models can be effective if their accuracy is higher and the
model outcome is closer to reality (Palasca, 2012). Regression is one of the
techniques used to determine the relationship between independent and
dependent variables (Hopkins & Ferguson, 2014; Hosmer Jr, Lemeshow, &
Sturdivant, 2013), which may be either linear or non-linear used for forecast-
ing (Bansal, Kauffman, & Weitz, 1993). A regression model is performed to
select the best independent variables along with their correlation and mag-
nitude that have a direct effect on the dependent variable (Larasati, DeYong,
& Slevitch, 2011). There is a difference between correlation that may be
 Addressing Regression for Business Applications ◾ 31

either positive or negative, but it does not indicate one is the cause for the
other and causality that is a confrmation of causal relationship based on the
causal theory (Rubinfeld, 2011). A hypothesis test can be performed to sub-
stantiate the proposed theory using cross-sectional analysis when the data
is a sample from the population at a specifc point of time and time series
analysis when the data is collected from different periods (Rubinfeld, 2011).
The discussion of different regression models is in Section 3.2 and the busi-
ness applications of the regression model will be discussed in Section 3.3.
The conclusion will be discussed in Section 3.4.

3.2 Popular Regression Models

3.2.1 Simple Linear Regression
Simple linear regression is the technique used to fnd the prediction and cor-
relation between an independent variable and a dependent variable (Bayaga,
2010; Bolton, 2009; Omay, 2010), which is also used in machine learn-
ing (ML) to get the output from the linear form of connections using input
variables (Buteneers, Caluwaerts, Dambre, Verstraeten, & Schrauwen, 2013).
The assumptions of linear regression are as follows: the independent and
dependent variables should have a linear relationship; a variance of residuals
should be constant; values (residual) should be independent; and the distri-
bution should follow the normal distribution (Bolton, 2009).

3.2.2 Multiple Linear Regression

Assumptions of multiple linear regression will be the same as simple lin-
ear regression (Omay, 2010) along with avoidance of multi-collinearity that
will increase the chance of type II error (Qasim, Månsson, & Kibria, 2021),
among the independent variables (Detienne, Detienne, & Joshi, 2003).
Incorporating all the dependent variables will give the model strength to
predict the result precisely; failing to do so will cause bias in the model
affecting the statistical result (Rubinfeld, 2011).

3.2.3 Logistic Regression

Logistic regression, which is an extension of logit transformation (Omay,
2010), is different from linear regression in terms of dependent variable
32 ◾ Machine Learning for Business Analytics

outcome (Hosmer Jr et al., 2013; Omay, 2010) that is either in binary form
or dichotomous form(Bolton, 2009). If the regression is linear regression,
the dependent variable outcome will be in continuous form. On the other
hand, logistic regression will have a discrete or categorical outcome for the
dependent variable (Hosmer Jr et al., 2013; Omay, 2010). Binomial logistic
regression is used in the case of two categorical outcomes for a dependent
variable, whereas more than two categorical outcomes for a dependent
variable can be handled by multi-nomial logistic regression (Bayaga, 2010).
Here, there is no need for the independent variables to be in the normal
distribution and for both variables to be in a linear relationship, paving
the way for more fexibility when compared to multiple linear regression
(Bayaga, 2010; Omay, 2010). An ordinal logistic regression model is preferred
when the dependent variable is in the ordinal form (rank order), which does
not help deal with multi-collinearity between independent variables (Larasati
et al., 2011).

3.2.4 Ridge Regression

Ridge regression, also known as a curative model of linear regression
(Kibria, Månsson, & Shukur, 2011), is useful in handling multi-collinearity
between the independent variables with the help of ridge parameter and
identity matrix and is given a higher performance in results than ordinary
least-squares method particularly in dealing with high correlation between
independent variables (Bager, Roman, Algelidh, & Mohammed, 2017). It
incorporates an extra element to the loss function, which is useful in tun-
ing the regularization parameter of ridge regression for eliminating the
overftting problem (Buteneers et al., 2013). Besides, a likelihood function is
improved by penalizing only the coeffcients of the independent variables
but not the intercepts (Pereira, Basto, & Silva, 2015). However, the ridge
regression with this penalty will make the coeffcients nearly close to zero
but not exactly zero creating a hindrance in selecting the predictors in the
model (Pereira et al., 2015).

3.2.5 Lasso Regression

The problem in decreasing the independent variables by the ridge regression
will be solved by the lasso regression as the introduced penalty in the lasso
regression acts not only to direct the coeffcients to zero but also to select
the correct independent variables in the model (Pereira et al., 2015).
 Addressing Regression for Business Applications ◾ 33

3.2.6 Random Forest Regression

Random forest is an amalgamation of various trees in which each tree sub-
set of the observation is chosen randomly (Grömping, 2009). Random for-
est regression is used for Boston housing price prediction (Liaw & Wiener,
2002). Because the overall prediction is based on the prediction of an indi-
vidual tree that delivers a multidimensional step function, the outcome of
the overall prediction will be smooth (Grömping, 2009).

3.2.7 Support Vector Regression

Support vector regression is based on learning theory (statistics), which is
used for the classifcation of objects as well as optimal characters (Smola
& Scholkopf, 2004). This gives the best result when compared to another
model like feature space representation on Boston house price estimation
(Drucker, Burges, Kaufman, Smola, & Vapnik, 1996).

3.2.8 Quantile Regression

Quantile regression, which is an extension of quantile function (conditional),
is used for the estimation of a conditional-based model (Koenker & Hallock,
2001). The word quantile means a sample is divided into equal sized and
sub groups, which means 25%, 50%, and 75% are the sample quantiles
(Yu, Lu, & Stander, 2003) that can be used to observe extreme values of the
samples (Jareño, Ferrer, & Miroslavova, 2016). Linear regression will use only
the average relationship between dependent and independent variables,
whereas the quantile regression will give a clear picture of the relationship
between both variables by plotting quantile regression curves (Yu et al.,
2003). The coeffcients obtained by quantile regression will be distributed
by the outliers (Jareño et al., 2016) due to the usage of a weighted sum of
absolute deviation, and the estimators of quantile regression will deliver an
effective performance than ordinary least square due to error terms that are
not in the normal distribution (Hung, Shang, & Wang, 2010).

3.2.9 Limitations of Regressions

Multiple linear regression has a severe drawback in handling the non-
linearity problem (Detienne et al., 2003). The sophistication of the regres-
sion model will be increased if the correlation is high between independent
34 ◾ Machine Learning for Business Analytics

variables, leading to the elimination of some important independent vari-

ables due to multi-collinearity and the addition of some other unwanted
independent variables (Pereira et al., 2015). The assumption made in mod-
eling the regression is based on the researcher’s choice that will lead to
affect the decision made from the regression results (Detienne et al., 2003).
Also, adjusting the regression line from the least square value will affect the
assumption from which the regression model is built (Bolton, 2009).
Besides, the elimination of coeffcients that is nearly equal to zero will
not give any contribution to the regression model, so the improvisation
of data is not there (Detienne et al., 2003). Overftting will be a problem
when the number of samples (rows) is lower than the total number of
independent variables (Pereira et al., 2015). If the variables are correlated
more strongly, even more data will not be helpful (Detienne et al., 2003).
The ridge regression has a disadvantage that standard errors lack, which
makes the predicted coeffcients cannot be tested for their signifcance
(Annaert, Claes, Ceuster, & Zhang, 2013). When it comes to outliers in
the data, linear regression is not effective in dealing with those outliers
(Detienne et al., 2003).

3.3 Applications of Regression Models in

Four Domains of the Business
3.3.1 Finance
3.3.1.1 Credit Score
Simple linear regression is used to fnd the credit score (Bolton, 2009)
that has been evaluated for not only individuals (customers) but also
businesses (Bensic, Sarlija, & Zekic-Susac, 2006). This credit score is
effectively useful to understand the potential of the borrower in terms of
repayment. Even if the credit score evaluation is for business, the owner
of the business will be evaluated for the repayment of borrowed loan
amount by measuring his/her credit score (Bensic et al., 2006) where the
logistic regression is used for deciding whether the credit risk is positive
or negative (Satheesh & Nagaraj, 2021). Bolton (2009) has collected data
from the bank that was used to form various clustering after eliminat-
ing the non-signifcant variables. Then, stepwise logistic regression was
used to increase the potential of the model in predicting the credit score
(Bolton, 2009).
 Addressing Regression for Business Applications ◾ 35

3.3.1.2 Risk Analysis

Awareness, identifcation, tracking, and palliation of risk are required to
develop and analyze the risk analysis method. A risk analysis model can
be built from the relationship between the independent and the dependent
variables. Multiple linear regression has given a better accuracy rate of 60.5%
in determining the likelihood of the risk with the help of signifcant predict-
ing variables (Bayaga, 2010).

3.3.1.3 Business Performance

There are two elements that aid in the evaluation of business performance:
internal behavior and external environment along with various organiza-
tional performance determinants like return on investment, profts earned,
size of the company, number of years since establishment, and its position
in the association (Wood, 2006). An ordinal logistic regression and a binary
logistic regression were used in the United Kingdom small frm data to fnd
the best model that would provide the major factors affecting the frm’s
performance. And the results showed that the number of active years, pro-
motion procedures, family funds, and struggle in hiring activities were the
major determinants that appeared to affect business performance. However,
the ordinal logistic regression was delivered a better performance in identi-
fying the signifcant variables than binary logistic regression (Wood, 2006).

3.3.1.4 Revenue Model

Additional services along with the core service will give more revenue to
the organization. For example, motivating passengers to engage in shopping
during the waiting time at the airport will increase the revenue for the orga-
nization and, at the same time, will improve the traveling experience for the
customers (Fasone, Kofer, & Scuderi, 2016). A ridge regression model and a
partial least-squares regression model are used to identify the factors in non-
aviation services urging to shopping, due to keeping all the independent vari-
ables that have a very high correlation with each other (Fasone et al., 2016).

3.3.1.5 Value at Risk

Value at risk model that uses quantile regression approach is an indicator
of market risk, which requires banks to inform on a daily basis to develop
36 ◾ Machine Learning for Business Analytics

fnancial regulations (Yu et al., 2003). Value at risk is based on the fnancial
return that is demonstrated by several quantiles so the quantile regression is
the best approach to evaluate it (Yu et al., 2003).

3.3.1.6 Business Failure

Business failure may happen due to various reasons, namely, insolvency,
legal problems, or consecutive losses (Pereira et al., 2015). Thus, it is essen-
tial to predict corporate failure in advance to make the right decision for
preventive measures. Pereira et al. (2015) were used ridge logistic and lasso
logistic regression to predict the business failure with another method
like the stepwise method, results of the ridge and lasso logistic regression
showed lower type I error than the stepwise method.

3.3.2 Marketing
3.3.2.1 Direct Marketing Score Model
Direct marketing is the way of communicating to the potential customers
with specifed offers by the marketing team to urge them to involve purchase.
However, it is not an easy task to do because each communication will cost
the company and, at the same time, not every customer is ready to purchase
immediately even if the offer is so attractive (Malthouse, 1999). Market segmen-
tation is one of the basic marketing practices to break down the entire market
customer into small market customer groups that will be helpful to apply right
strategies for enhancing the relationship between the organization and the cus-
tomers (Chattopadhyay, Dan, Majumdar, & Chakraborty, 2012). Even after seg-
menting, the targeting should be selected based on the customer’s willingness
to make a purchase, which can be done using directing marketing score model
that has so many explanatory variables (Malthouse, 1999). The ridge regression
is used to analyze the performance of direct marketing score because many
variables can be included in the model without worry about the multi-collin-
earity and overftting assisting marketers to precisely understand the impact of
predictive variables on direct marketing (Malthouse, 1999).

3.3.2.2 Family Business Research

The performance of the family business and family ownership handover and
the relationship between the involvement of family members in the board and
global sales were evaluated with regression analysis (Hopkins & Ferguson,
 Addressing Regression for Business Applications ◾ 37

2014). Moreover, the relationship between ownership of founding family and

business performance was also evaluated by multiple linear regression analysis
due to the involvement of many independent variables, namely the number of
family members involved, incentives of family and outside members, the num-
ber of years served in top management positions, and the number of outside
members in the board (Hopkins & Ferguson, 2014).

3.3.2.3 Market Consumption and Price Determinants

Accessing individuals to do the study of the market will be diffcult, which
leads to group formation. Those groups can be divided into high, medium, and
low groups and can be studied with the help of quantile regression (Yu et al.,
2003). Pricing of a hotel room is infuenced by so many factors like marketing
tools involved and its strategies, room rates, demand in the market, consumer
perception, customer satisfaction, which can be solved and analyzed by quan-
tile regression to get the signifcant pricing determinants (Hung et al., 2010).
Besides, multiple linear regression is used to predict the price of the apartments
that have many explanatory variables (Ceh, Kilibarda, Lisec, & Bajat, 2018).

3.3.2.4 Business Growth

Although several incentives are offered by the US government to the US
farmers, business strategies—namely, cost reduction, revenue improvement,
asset, and fnancial management—are deployed for maintaining a stable
income and equity growth. Those business strategies that are performed
in different quantiles can be examined using quantile regression analysis
(Hennings & Katchova, 2005). Also, the infuence of acquisition performance
on the level of company distress and the relationship between monitoring
rating and performance and quality along with the duration of work are
performed with the help of regression analysis (Aguinis, 1995).

3.3.2.5 Export Market

A company involved in exporting to foreign countries will not have the same
resources used by the other export company due to the difference between
companies with respect to frm size, availability of skilled labor, research
and development facility, etc. (Wagner, 2004). Several organizational hurdles
prevent successful exports—namely, poor knowledge of international mar-
kets, lack of foreign cultural exposure, and inadequate foreign networks
(Bennett, 1997). To fnd out the relationship between export sales ratio and
plant attributes, Wagner (2004) has used the quantile regression.
38 ◾ Machine Learning for Business Analytics

3.3.3 Human Resource Management

3.3.3.1 Organizational Justice and Employee Commitment
Organizational justice is referred to fairness in decision-making and its
perception in organizations, which affects the employee commitment that
has a direct connection to the performance of the employees (Jamaludin,
2009). It is classifed into three justice: the frst one is distributive justice
that is perceived understanding of resource distribution, the second one
is procedural justice that is related to the compensation assigned to the
workload, and the third one is interactional justice that is the interaction
between the employees and the organization. There are three types of
commitments: an affective commitment that is the emotional and psycho-
logical attachment of the employee with the company, normative commit-
ment that is a feeling of responsibility to the organization, and continuance
commitment that is a cost-associated consideration of leaving the organiza-
tion combined with the opportunities outside the organization (Jamaludin,
2009). Jamaludin (2009) has conducted a linear regression model to under-
stand the infuence of organizational justice on employee commitment
and concluded that interactional justice has a strong impact on affective
commitment and distributive justice has no impact on any of the above-
mentioned commitments.

3.3.3.2 Employee Selection

Hiring a candidate is very critical in the services industry, where the per-
formance of the employee plays a major role in business success. Many
variables are available in the database to be considered to select the right
candidate, which has to cluster into four groups: personal variables that
include personal details, test variables that include scores of exams and
interview, occupational variables that include employment history, and per-
formance variables that include promotion and career score (Sebt & Yousef,
2015). Sebt and Yousef (2015) have used stepwise regression analysis to
obtain the best predictive variables for the employee selection after perform-
ing a decision tree in the data mining process and have concluded that the
fve best variables are identifed from the total of 26 variables.

3.3.3.3 Human Capital Development

To survive in the competitive business world, the organization should have
the competitive advantage that can be formed by enhancing the skills of
 Addressing Regression for Business Applications ◾ 39

employees by providing proper training and opportunity (Florea & Mihai,

2015). It is a win–win situation for both employees and the organization to
allow the employees for upskilling. Florea and Mihai (2015) have used multi-
variate regression to address the relationship between the number of training
hours and the number of employees undergoing that training and the organiza-
tional performance and concluded that the number of training hours has a pos-
itive impact on the organizational performance without hiring more employees.

3.3.3.4 Payroll and Compensation

In labor market, salary expectation relies on how productive the labor is
and how many years of experience the labor has got. The multiple regres-
sion is used to fnd out salary discrimination based on sex and other
variables—namely, the level of education and the number of years in the
job. While selecting the dependent variable for salary in the model, choos-
ing the hourly wage is good if the number of work hours is based on the
work preference alone and the selection of annual salary is good if there is
any discrimination (Rubinfeld, 2011).

3.3.4 Operations
3.3.4.1 Demand Forecasting
When the manufacturing company perceives the demand of the product based
on the customer preference, the production and the inventory can be main-
tained effectively (Guanghui, 2012) along with effcient supply chain collabora-
tion (Carbonneau, Laframboise, & Vahidov, 2008). Not only predicting uncertain
demand but also fulflling that demand with various existing constraints such as
proper production schedule and its timely execution, reduction of time between
the order and the delivery, and multi-team coordination issues is a challeng-
ing task (Guanghui, 2012). Even if the demand is predicted correctly, the com-
pany demand will deviate from the actual demand due to the bullwhip effect
(Carbonneau et al., 2008). Guanghui (2012) has preferred the support vector
regression to forecast the supply chain demand of dairy products and has come
to a conclusion that support vector regression gives higher accuracy.

3.3.4.2 Supply Chain Flexibility

Due to climate change and competitiveness, companies have to apply
new strategies to handle the supply chain partners because of multiple
40 ◾ Machine Learning for Business Analytics

dimensions involved in the process of the supply chain, namely, the capac-
ity of the company, product delivery, transaction of information with each
other, integration with plant and supply chain, and logistics (Jeeva & Guo,
2010). The multi-variate regression analysis is used by Jeeva and Guo (2010)
to study the supply chain fexibility, and they concluded that multi-variate
regression will be more effective if the number of suppliers is limited.

3.3.4.3 Green Supply Chain Management

Though environmental pollution has been a series issue for a few decades,
industrialization is inevitable for country growth paving the way for green
supply chain management that drives the organization to enhance organiza-
tion environmental performance. Various hypotheses, namely, internal and
external practices of green supply chain management on organization fex-
ibility, environment pollution, and operational cost are formed and studied
using linear regression (Mumtaz, Ali, & Petrillo, 2018).

3.4 Conclusion
The business has been incorporating most of the statistical methods along
with state-of-the-technology for many decades. Starting from Microsoft Excel
to the advanced technologies uses the regression analysis for identifying the
relationship between variables. It is advisable to perform regression in the
frst place before proceeding to any complex model because regression anal-
ysis has been a robust statistical technique for years. Hence, several popular
models of regression and its applications are discussed with business cases.
Though there are new methods introduced in the artifcial intelligence era,
regression has its advantages.

References
Aguinis, H. (1995). Statistical power with moderated multiple regression in man-
agement research. Journal of Management, 21(6), 1141–1158.
Annaert, J., Claes, A. G. P., Ceuster, M. J. K. D., & Zhang, H. (2013). Estimating the
spot rate curve using the Nelson-Siegel model: A ridge regression approach.
International Review of Economics and Finance, 1–15. doi:10.1016/j.
iref.2013.01.005
 Addressing Regression for Business Applications ◾ 41

Bager, A., Roman, M., Algelidh, M., & Mohammed, B. (2017). Addressing multicol-
linearity in regression models: A ridge regression application. Munich Personal
RePEc Archive, 1–20.
Bansal, A., Kauffman, R. J., & Weitz, R. (1993). Comparing the performance of
regression and neural networks as data quality varies: A business value
approach. Journal of Management Information Systems, 11–32. doi:10.1080/07
421222.1993.11517988
Bayaga, A. (2010). Multinomial logistic regression: Usage and application in risk
analysis. Journal of Applied Quantitative Methods, 5(2), 288–297.
Bennett, R. (1997). Export marketing and the Interne. International Marketing
Review, 14(5), 300–323. doi:10.1108/02651339710184280
Bensic, M., Sarlija, N., & Zekic-Susac, M. (2006). Modeling small business credit
scoring by using logistic regression, neural networks, and decision trees.
Intelligent Systems in Accounting, Finance and Management, 13(3), 133–150.
doi:10.1002/isaf.261
Bergeron, F., Raymond, L., & Rivard, S. (2004). Ideal patterns of strategic alignment
and business performance. Informations and Management, 41, 1003–1020.
Bolton, C. (2009). Logistic regression and its application in credit scoring
[Unpublished Master’s thesis]. University of Pretoria, 1–240.
Buteneers, P., Caluwaerts, K., Dambre, J., Verstraeten, D., & Schrauwen, B. (2013).
Optimized parameter search for large datasets of the regularization parameter
and feature selection for ridge regression. Neural Processing Letters, 38, 403–416.
Carbonneau, R., Laframboise, K., & Vahidov, R. (2008). Application of machine
learning techniques for supply chain demand forecasting. European Journal of
Operational Research, 184, 1140–1154. doi:10.1016/j.ejor.2006.12.004
Ceh, M., Kilibarda, M., Lisec, A., & Bajat, B. (2018). Estimating the performance of
random forest versus multiple regression for predicting prices of the apartments.
International Journal of Geo-Information, 7(168), 1–16. doi:10.3390/ijgi7050168
Chattopadhyay, M., Dan, P. K., Majumdar, S., & Chakraborty, P. S. (2012).
Application of neural network in market segmentation: A review on
recent trends. Management Science Letters, 2(2), 425–438. doi:10.5267/j.
msl.2012.01.013
Detienne, K. B., Detienne, D. H., & Joshi, S. A. (2003). Neural networks as statis-
tical tools for business researchers. Organizational Research Methods, 6(2),
236–265. doi:10.1177/1094428103251907
Drucker, H., Burges, C. J. C., Kaufman, L., Smola, A., & Vapnik, V. (1996). Support
vector regression machines. NIPS’96: Proceedings of the 9th International
Conference on Neural Information Processing Systems, 9, 155–161.
Fasone, V., Kofer, L., & Scuderi, R. (2016). Business performance of airports:
Non-aviation revenues and their determinants. Journal of Air Transport
Management, 53, 35–45. doi:10.1016/j.jairtraman.2015.12.012
Florea, N. V., & Mihai, D. C. (2015). Improving organization performance through
human capital development, using a regression function and Matlab. Journal
of Science and Arts, 15(32), 229–238.
42 ◾ Machine Learning for Business Analytics

Grömping, U. (2009). Variable importance assessment in regression: Linear

regression versus random forest. The American Statistician, 63(4), 308–319.
doi:10.1198/tast.2009.08199
Guanghui, W. (2012). Demand forecasting of supply chain based on support vec-
tor regression method. Procedia Engineering, 29, 280–284. doi:10.1016/j.
proeng.2011.12.707
Hennings, E., & Katchova, A. L. (2005). Business growth strategies of Illinois
farms: A quantile regression approach. 2005 Annual Meeting, July 24–27,
Providence, RI from American Agricultural Economics Association (New Name
2008: Agricultural and Applied Economics Association), 1–29. doi:10.22004/
ag.econ.19367
Hopkins, L., & Ferguson, K. E. (2014). Looking forward: The role of multiple
regression in family business research. Journal of Family Business Strategy, 5,
52–62. doi:10.1016/j.jfbs.2014.01.008
Hosmer Jr, D. W., Lemeshow, S., & Sturdivant, R. X. (2013). Applied logistic regres-
sion (3rd ed.). Wiley.
Hung, W.-T., Shang, J.-K., & Wang, F.-C. (2010). Pricing determinants in the hotel
industry: Quantile regression analysis. International Journal of Hospitality
Management, 29, 378–384. doi:10.1016/j.ijhm.2009.09.001
Jamaludin, Z. (2009). Perceived organizational justice and its impact to the
development of commitments: A regression analysis. World Journal of
Management, 1(1), 49–61.
Jareño, F., Ferrer, R., & Miroslavova, S. (2016). US stock market sensitivity to inter-
est and infation rates: A quantile regression approach. Applied Economics,
1–13. doi:10.1080/00036846.2015.1122735
Jeeva, A. S., & Guo, W. W. (2010). Supply chain fexibility assessment by multivari-
ate regression and neural networks. Advances in Neural Network Research
and Applications, 67, 845–852. doi:10.1007/978-3-642-12990-2_98
Kibria, B. M. G., Månsson, K., & Shukur, G. (2011). Performance of some logistic
ridge regression estimators. Computational Economics, 401–414. doi:10.1007/
s10614-011-9275-x
Koenker, R., & Hallock, K. F. (2001). Quantile regression. Journal of Economic
Perspectives, 15(4), 143–156.
Larasati, A., DeYong, C., & Slevitch, L. (2011). Comparing neural network and
ordinal logistic regression to analyze attitude responses. Service Science, 3(4),
304–312. doi:10.1287/serv.3.4.304
Liaw, A., & Wiener, M. (2002). Classifcation and regression by random forest. R
News, 18–22.
Malthouse, E. C. (1999). Ridge regression and direct marketing scoring
models. Journal of Interactive Marketing, 13(3), 10–23. doi:10.1002/
(SICI)1520-6653(199923)13:43.0.CO;2–3
 Addressing Regression for Business Applications ◾ 43

Mumtaz, U., Ali, Y., & Petrillo, A. (2018). A linear regression approach to evalu-
ate the green supply chain management impact on industrial organizational
performance. Science of the Total Environment, 624, 162–169. doi:10.1016/j.
scitotenv.2017.12.089
Omay, C. (2010). Logistic regression: Concept and application. Educational
Sciences: Theory and Practice, 10(3), 1397–1407.
Palasca, S. (2012). Emerging markets queries in fnance and business statistical
evaluations of business cycle phases. Procedia Economics and Finance, 3,
119–124. doi:10.1016/S2212-5671(12)00129-3
Pereira, J. M., Basto, M., & Silva, A. F. d. (2015). The logistic lasso and ridge regres-
sion in predicting corporate failure. Procedia Economics and Finance, 39,
634–641. doi:10.1016/S2212-5671(16)30310-0
Qasim, M., Månsson, K., & Kibria, B. M. G. (2021). On some beta ridge regres-
sion estimators: Method, simulation and application. Journal of Statistical
Computation and Simulation, 1–14. doi:10.1080/00949655.2020.1867549
Rubinfeld, D. L. (2011). Reference guide on multiple regression. In Reference
manual on scientifc evidence (3rd ed., pp. 303–359). Washington, DC: The
National Academies Press.
Satheesh, M. K., & Nagaraj, S. (2021). Applications of artifcial intelligence on
customer experience and service quality of the banking sector. International
Management Review, 17(1), 9–17.
Sebt, M. V., & Yousef, H. (2015). Comparing data mining approach and regression
method in determining factors affecting the selection of human resources. Fen
Bilimleri Dergisi (CFD), 36(4), 1846–1859.
Smola, A. J., & Scholkopf, B. (2004). A tutorial on support vector regression.
Statistics and Computing, 14, 199–222.
Sofe, B., & Hubert, O. (2004). 35 years of studies on business failure: An over-
view of the classic statistical methodologies and their related problems. Vlerick
Leuven Gent Working Paper Series, 1–70.
Wagner, J. (2004). Export intensity and plant characteristics: What can we learn
from quantile regression? HWWA Discussion Paper, 304, 1–6.
Wood, E. H. (2006). The internal predictors of business performance in small
frms a logistic regression analysis. Journal of Small Business and Enterprise
Development, 13(3), 441–453. doi:10.1108/14626000610680299
Yu, K., Lu, Z., & Stander, J. (2003). Quantile regression: Applications and current
research areas. The Statistician, 52, 331–350.
Zuo, Y., Kajikawa, Y., & Mori, J. (2016). Extraction of business relationships in sup-
ply chain networks using statistical learning theory. Heliyon, 2(6), e00123.
Chapter 4

Chatbots: Their Uses

and Impact in the
Hospitality Sector
Princy Sera Rajan, Darsana S. Babu, and Sameena M. H.

Contents
4.1 Introduction ..............................................................................................45
4.2 Literature Review .....................................................................................46
4.3 Proposed Methodology ............................................................................47
4.4 Limitations of Using Chatbots..................................................................53
4.5 Conclusion ................................................................................................54
4.6 Future Scope ............................................................................................55
References .........................................................................................................55

4.1 Introduction
In recent times, there is an increasing need for improvement in any industry
to keep up with the customer’s expectations. Especially in the hospitality
industry, as it is a service industry, it deals with the customer service experi-
ence, which directly aligns with their satisfaction to be in line and acts as
fast as possible; the easiest and immediate solution would be the integration
of automated technology using artifcial intelligence (AI)-driven chatbots.
AI is the realm of computer science that focuses on the functions of
intelligent machines; they mimic human activities and reactions [1]. Some

DOI: 10.4324/9781003206316-4 45
46 ◾ Machine Learning for Business Analytics

of the AI-enabled functions include speech recognition, learning, planning,

and problem-solving. The conversation is an intelligent device in the AI
spectrum, for which it is designed for the specifc purpose to meet its user
requests. Conversations are understood exactly what the user entries for
different details, such as focusing on specifc keywords from a conversation
between a bot and an individual user [1]. It requires planning and comple-
tion of work before the chatbot can move on to its next task and also iden-
tifying the fle sequence of actions to complete its main tasks there when it
comes to complex tasks [2].
Chatbots offer a few key benefits. The first advantage is that the
chatbot is available 24/7 to offer App. Another advantage is that the
customer waiting time is removed, so customers can access the service
any time. For example, in the hospitality context, customers do not
have to wait in line to get basic works in a booking room. Chatbots
give them the virtual representation of available options and their
values and help them make reservations. Another advantage is custom-
ization. Interviews work as existing helpers who can keep customer
details with the agent continuously so that the operator can provide
the client with better information as well as solutions based on current
needs and previous discussions [3].
Therefore, they help hotels to customize the customer experience in all
stages of the tourist cycle, from the pre-arrival stage to the post-departure
phase. This increases product reliability as well as opportunities to visit
again and again. Chatbots can also help manage internet traffc and make
sure customers get the right kind of information and assistance to improve
conversion and reduce stroke. This is very important for businesses that rely
heavily on online marketing.
Chatbots can also be used to build customer relationships. For example,
a chatbot may be designed to please the customer on his birthday and post
promotional offers to visit places. Such actions allow businesses to develop
customer experiences while performing marketing activities [4].

4.2 Literature Review

In today’s modern generation of service providing, customers expect fast
and quality service, which can be provided by using AI. The chatbot is
like an application designed to make a conversation with the human.
 Chatbots: In Hospitality Sector ◾ 47

They are designed to understand the requirement of the person they are
having a conversation with and answer with appropriate messages and
give recommendations according to their preference. The usage of a chat-
bot or AI in the feld of hospitality is more proftable and cost-effective
than having a human agent to address the customers [1]. Chatbots, in
the hospitality sector, are involved in hotels’ applications for the ease of
use with mobile phones; this helps them increase their online presence
in social media as well as reach more target customers. They could hold
conversations and can communicate with customers from different regions
in their language [2].
Even though chatbots have their pitfalls, they have adapted rapidly in the
whole hospitality industry from hotel room booking, to enabling services
with just a click, to fight bookings [1].

4.3 Proposed Methodology

Considering the methodology for the four research papers mentioned, the
following are the inputs, discussions, and the outputs that drive the papers
through in and throughout the impact of AI or the use of chatbots in the
hospitality industry is shown in Table 4.1.
The percentage of respondents’ preference on chatbots from Table 4.2 is
shown in Figure 4.1.
The percentage of respondents’ feedback on customer satisfaction
through chatbot from Table 4.3 is shown in Figure 4.2.
The percentage of respondents’ age reacting to AI from Table 4.4 is
shown in Figure 4.3.
The percentage of respondents’ awareness towards chatbots from Table
4.5 is shown in Figure 4.4.

Concluding Methodology: Having listed the inputs, discussions, and the

outputs derived out of all the papers individually, the common objec-
tives of the papers like fnding the use of chatbots in the hospitality
industry, leading the results to an increased level of customer satisfac-
tion and heavy technology features in the hospitality industry, and
taking the baby steps through the chatbots have completed the purpose
and the objective of the study through effective input utilization and
effcient discussions in Figures 4.1–4.4.
 48
Table 4.1 Different Methodologies

◾
Paper Inputs Considered Discussions Resultant Outputs

Machine Learning for Business Analytics

Reference
Research With about 100 With effcient analysis of They found the outputs as the following, Age outputs:
Paper 1: respondents, the paper the data responses 13% of the respondents were aware of AI; 65% of the
An formed a questionnaire, received, they respondents had a vague idea of AI; 22% of the
explanatory with all the inputs differentiated the age, respondents were not at all aware of AI. Age group of
study of required for the survey awareness, usage, 25–40 years had a vague idea about AI. Preferences
customer for arriving at the desired preferability, and all the output: 10% of the respondents were not open to being
perceptions outputs. required base factors for served by either; 16% of the respondents were not sure;
of usage of The questionnaire was: 1. the study of chatbots in 69% of the respondents preferred a real person for
chatbots Are you aware of the the hospitality sector. availing of customer service; 4% preferred a virtual
in the chatbot feature leading The main discussion was assistant; and 12% preferred both is shown in Table 4.2
hospitality hotels have deployed? about the chatbot Feedbacks received on customer satisfaction: 15% said that
sector [1]. Have you used it? injecting any extra it depended on the situation; 34% of the respondents
customer satisfaction felt that technology played an important role in the level
2. To what extent do you
throughout their stay in of satisfaction with the service; 9% of the respondents
know about artifcial
any hotel. And their thought that a combination of technology and personal
intelligence (AI)?
outputs are discussed touch was required; 29% thought that it Thought that it
3. While staying at a hotel, elaborately. depended on the service context shown in Table 4.3
would you like to be
served by robots and/or
chatbots from check-in to
check-out?
Research The inputs were gathered The study has used the The plotting of the user inputs and the recognition of
Paper 2: by the following sources: NLU, NLG, and NLP for various features of the language is done through the
Smart hotel 1. NLU—natural language driving the inputs into NLU, while the NLG is used for text planning (for related
using understanding guiding the chatbot and relevant chatbot replies), sentence planning
intelligent 2. NLG—natural language through effcient use of (grammatical formation and the replacing with the right
chatbot [5]. generation AI for guiding the choice of words, phrases, and the fundamental attitude
chatbots for use. and behavior in the conversation, i.e, on subtle
3. NLP—natural language
Customer satisfaction is grounds), text realizations (structuring out the sentence
processing
also a bonus beneft for on the basis of the aforementioned factors), and NLP is
4. Data collection through used for computerized formats (by gathering the
fnding out the uses that
questionnaire. knowledge on how the humans act on various situations
chatbots give toward
customer satisfaction. and answer on various questions through tools, codes,
The data source of 150 program, techniques that are designed in the chatbots
respondents formed the based on the aforementioned document analysis).
output part of the study. Results of the questionnaire also drive the output results
as follows: The 150 sampled questionnaire showed results
The questionnaire is also
of the following consumer feedback results: 40% of the
supported for the sake of

Chatbots: In Hospitality Sector

respondents were already satisfed with the provided
fnding customer
privacy, safety, and the security of the hotel they
satisfaction. The studies
checked in. While 20% were not satisfed, the reason
also provide the art of
mentioned differed. They specifed the location of the
making its services more
hotel is the reason for their decision, while 40% felt that
unique, effcient, and
their consumer experience can adapt to the new grown
effective for the users
technologies and other facilities provided to the
and the readers.
customers who checked in the latest, through proper
awareness about the hospitality industry and the
awareness about the use of AI in the same is discussed
in Tables 4.4 and 4.5
(Continued)

◾ 49
Table 4.1 (Continued)

50 ◾
Paper Inputs Considered Discussions Resultant Outputs
Reference

Machine Learning for Business Analytics

Research The paper has done its Through qualitative data Through pre-programmed conversations and suggesting
Paper 3: research about the use of being sieged and places that the user might want to explore, the chatbot
Chatbots—an chatbots in the hospitality provided, the inputs are provides all the details by understanding and responding.
organization’s industry. The authors also incurred in the effects By customizing all the languages, chatbots are the
friend or foe? discuss the rapidity it and are also categorized surprising advantages for all the hotels using AI as their
[2]. causes in various sectors into benefts and pain tool of development. Thus, AI is capturing its steps into the
and has also provided points. With the use of hospitality industry through chatbots is discussed in Tables
ample factors for chatbots, the paper 4.4 and 4.5 Chatbots can manage numerous channels and
increasing its impact in shows that the customer their customers at the same time. The Internet, being used
theindustry. The inputs experience has for numerous reasons, is now effcient through chatbots
have also been improved, thus the through travel agency apps like MakeMyTrip, GoIbibo.
improvised through hotels in the hospitality Thus, through a signature move, the hotels are just a click
effcient resource sector are developing away from the customer. Still, chatbots come with a
utilization. themselves in the package that includes limitations like privacy, less
technology or the AI intuitiveness, less awareness, and the chances of
aspects. consumers building an uncomfortable feeling.
Research 1. Understand the factors We adopted two core Based on isomorphic pressures, learning capabilities,
Paper 4: infuencing the frm-level organizational theories: learning competencies, and barriers are the control
Chatbot adoption of chatbots by institutional theory and variables and adoption intention is the dependent
adoption in tourism and hospitality organizational learning variable.
tourism industries. theory. Almost 85% of all interactions will be carried out
services: A 2. Examining the role of through chatbots in future.
conceptual chatbots in various areas
exploration of the tourism and
[3]. hospitality industry.

Note: The methodology of the Research Paper [1] (an explanatory study of customer perceptions of usage of chatbots in the hospi-
tality sector) has also dipped an information of customer awareness of within the hospitality industry. The summarized out-
puts of the research paper are fguratively explained below.
 Chatbots: In Hospitality Sector ◾ 51

Table 4.2 Percentage of Respondents’ Preference on Chatbots

1 10% The respondents were not open to be served by either

2 16% The respondents were not sure

3 69% The respondents preferred a real person for availing customer service

4 4% Preferred a virtual assistant

5 12% Preferred both

Figure 4.1 Percentage of respondents’ preference on chatbots.

Table 4.3 Percentage of Respondents’ Feedback on Customer Satisfaction

through Chatbot/AI

1 15% That it depended on the situation.

2 34% The respondents felt that technology played an important role in
the level of satisfaction of the service.
3 9% The respondents thought that a combination of technology and
personal touch was required
4 29% Thought that it depended on the service context.
5 28% Thought that technology played no role in the level of satisfaction.
52 ◾ Machine Learning for Business Analytics

Figure 4.2 Percentage of respondents’ feedback on customer satisfaction through

chatbot/AI.

Table 4.4 Percentage of Respondents’ Age Reacting to AI

1 13% The respondents were aware of AI.

2 65% The respondents had a vague idea of AI.
3 22% The respondents were not at all aware of AI.

Figure 4.3 Percentage of respondents’ age reacting to AI.

 Chatbots: In Hospitality Sector ◾ 53

Table 4.5 Percentage of Respondents Awareness Toward Chatbots

1 6% The respondents knew about chatbots and had used them.

2 60% The respondents had not even heard of chatbots.
3 4% The respondents had heard of them but were not sure what they
were.
4 30% The respondents knew about chatbots but had never used them.

Figure 4.4 Percentage of respondents’ awareness toward chatbots.

4.4 Limitations of Using Chatbots

Using chatbots can have many advantages. Collectively, there are many limi-
tations described below.

◾ As chatbots are being designed using neural networks and machine

learning (ML), it will be impossible in some cases to give a response to
critical customer queries.
◾ Decision-making skills for chatbots are very low.
◾ These chatbots are very least ranked in customer retention.
54 ◾ Machine Learning for Business Analytics

◾ In this crucial and important sector like hospitality, it is very important

to handle a customer with convincing comments and answers.
◾ Chatbots are very effective in reducing manpower such as helpdesk and
reception.
◾ So, chatbots are usually programmed to answer for a certain set of
questions using ML but if a customer raises queries for any other ques-
tion rather than trained questions, it may give the same answer that
does not match.
◾ Therefore, chatbots are lagging in this decision and critical thinking
skills.

This book speaks about the usage of chatbots in the hospitality sec-
tor. These assumptions are being made by taking the tourist sector as the
key factor. We analyzed how customers think of aware chatbots. We took
respondents’ opinions on chatbots like how often they use chatbots; how
they are about these automated answering algorithms; and which ages more
frequently use these chatbots. This chapter fnally speaks about customer
satisfaction and its advantages in mere future.

4.5 Conclusion
Technology and innovation are in a fast-running phase in the present world.
Chatbots can be very advantageous in the hospitality sector for giving sug-
gestions within the hand without asking anyone in person. This reduces
the time and effort of an individual to make decisions at an instant and
proceed. These chatbots give suggestions based on ML and neural network
algorithms. These applications can give real-time solutions from the previ-
ously trained data sets and feedback. This helps in giving more accurate
suggestions and predictions than humans do. When we speak for the tour-
ism sector, it is very important for fnding true visiting spots and locations.
In this application, chatbots can be very useful in answering the user inputs.
Although there are limitations like decision-making skills and critical think-
ing skills, they can be corrected by giving updates to these chatbot applica-
tions or websites.
 Chatbots: In Hospitality Sector ◾ 55

4.6 Future Scope

As this analysis speaks about only four researches, it can be made better by
adding more research articles into consideration. Further work can be done
in analyzing how to increase decision-making and risk-taking skills in AI
chatbots. However, analyzing this particular topic requires more time and
effort.

References
[1] Mihir Dash & Suprabha Bakshi, An Exploratory Study of Customer Perceptions
of Usage of Chatbots in the Hospitality Industry, International Journal on
Customer Relations, 7 (2), 2019, 27–33.
[2] Emma Carter & Charlotte Knol, Chatbots—an organisation’s friend or foe?,
Research in Hospitality Management, 9 (2), 2019, Stenden Hotel Management
School, NHL Stenden University of Applied Sciences, 11.
[3] Dandison Ukpabi, Bilal Aslam & Heikki Karjaluoto, Chatbot Adoption in
Tourism Services: A Conceptual Exploration. In Stanislav Ivanov & Craig
Webster (Eds.), Robots, Artifcial Intelligence, and Service Automation in
Travel, Tourism and Hospitality. Emerald Publishing Limited, 2019, 105–121.
DOI: 10.1108/978-1-78756-687-320191006.
[4] Santosh Bisoi, Mou Roy & Dr. Ansuman Samal, Impact of Artifcial Intelligence
in the Hospitality Industry, International Journal of Advanced Science and
Technology, 29 (5), 2020, 4265–4276.
[5] Shubham Parmar, Megha Meshram, Parth Parmar, Meet Patel & Payal Desai,
Smart Hotel Using Intelligent Chatbot: A Review, International Journal
of Scientifc Research in Computer Science, Engineering and Information
Technology (IJSRCSEIT), 5 (2), March–April 2019, 823–829, ISSN: 2456–3307.
https://doi.org/10.32628/CSEIT11952246. Journal URL: http://ijsrcseit.com/
CSEIT11952246
Chapter 5

Traversing through the Use

of Robotics in the Medical
Industry: Outlining Emerging
Trends and Perspectives
for Future Growth
Gaurav Nagpal, Kshitiz Sinha, Himanshu Seth,
and Namita Ruparel

Contents
5.1 Introduction ..............................................................................................58
5.2 Evolution of Robotic Surgery ...................................................................60
5.3 Benefts of RAS ........................................................................................60
5.4 Current Market Scenario ..........................................................................61
5.5 Future Growth Estimates .........................................................................65
5.5.1 Reasons for the Lower Than Required Presence of
Surgical Robots .............................................................................66
5.6 Limitations of Robotics ............................................................................ 67
5.7 Future Directions for Increasing the Acceptance of
Robotic Surgeries......................................................................................68
5.8 Conclusion ................................................................................................69
References .........................................................................................................70

DOI: 10.4324/9781003206316-5 57
58 ◾ Machine Learning for Business Analytics

5.1 Introduction
Humans and artifcial intelligence have joined forces in numerous felds with
machines [1]. Robots have become the reality of the contemporary world,
enabling humans to become more effective and connected [2]. There are
many verticals in which robots are being used, having specifc purpose and
utility in each of them, and designed accordingly to assist the humans wher-
ever possible with higher accuracy and precision. The technology has been
increasing exponentially since the past decade. Every industry is opting for
inducing its average workfow with robotic technology as it has proven to
be benefcial not only in terms of proft but also in terms of reduced human
error and increased dexterity and accuracy. Also, this is the age of AI with
increasing automation and decreasing human intervention in the processes.
The amount of capital being invested in AI by the companies, as well as the
number of people studying AI, has witnessed an increasing trend.
Mentoring and on-the-job training in the feld of medical surgeries can
also be done with the help of robots the way it is being resorted to in the
other industries [3]. Any job that has an outcome which can be pre-stated or
guessed should ideally be done by machines so that the humans can have
the bandwidth to perform the open-ended jobs [4]. A risk-oriented frame-
work has been proposed by Dhar (2015) to evaluate which decisions should
be made by robots and which ones by humans [5].
The healthcare sector is one of the felds in which advancement is far
from saturation. New directions of research are being discovered every year
that calls for endless innovations. In this chapter, the authors describe the
prospects of robotic technology being used in the healthcare sector, specif-
cally focusing on surgical robots, the development of which is in a nascent
phase. The need for robotic surgery will eventually prevail because, with
the increasing population and medical awareness, the number of surgeries
is also expected to increase, leading to more demand for surgeons. Here,
robotic surgery can be a solution amidst the lower per-capita availability of
skilled medicos, as the operation time and burden on the surgeon will both
be reduced with robotic surgeries. Not only this, the recovery after a robotic
operation is proven to be faster than open surgery, leading to shorter hospital
stays which means more rooms for patients and lesser hospitalization costs.
Robotics surgery is defned as the involvement of robotically assisted
surgical (RAS) devices that enable the surgeon to use computer and soft-
ware technology to control and move surgical instruments through one or
more tiny incisions in the patient’s body (minimally invasive) for a variety
 Robotics in the Medical Industry ◾ 59

Figure 5.1 Classifcation of robots in the healthcare industry.

Source: Self-composed by the authors.

of surgical procedures. RAS devices generally have three main components:

a console, the bedside cart, and a separate cart containing the supporting
hardware and software [6]. The healthcare industry is undergoing automa-
tion with the rising number of robotic technologies. Various types of robots
are being used at present, each one having its unique features and applica-
tions in surgery. The most basic classifcation is shown in Figure 5.1.
Each classifcation in this fgure has a distinct approach to surgeries.
Surgical robots are used for performing normal operations on humans such
as general, laparoscopic, brain, spine, ocular, and cardiovascular which
involve cutting or probing specifc parts (electrodes and knee implants).
Navigational robots are used only for diagnosis and scanning of various
body parts for any pharmacological action, used in endoscopy. Three-
dimensional (3-D) and high-defnition (HD) vision can be obtained by the
robotic arm, used to determine if the operation is required for the individual
or not. Rehabilitation robots are robots that help in improving the body
movements of people who were paralyzed or had impaired body move-
ments. Robots can provide automatic body movements which can aid the
elderly and disabled. Skin modifcation robots are also available which are
used to obtain wrinkle-free and rejuvenated skin. The category “Others”
comprises hygiene maintaining robots that are not directly related to human
health but are a necessity in hospitals for maintaining sanitation, collec-
tion of records through face recognition, dispensing of medicines, and in
many other operations. Each type of surgery requires different techniques
to operate with and has a different set of patents. Various technologies have
been covered with innovations that are being developed in this industry.
Following the trend, major companies have started to invest in this sector.
Numerous procedures need to be followed even after a robot has been
developed and tested, before it is available to the population. As stated
earlier, surgeries of different parts of the body require a different technol-
ogy to operate with and have a different set of patents. Newly developed
60 ◾ Machine Learning for Business Analytics

innovations and discrete technologies in robotic healthcare have been cov-

ered in this research study. An aspect that is important to understand while
studying surgical robotics is patent, as it safeguards the invention of the
company and allows the inventor to enjoy market exclusivity throughout its
operational time and also is a major criterion when distinguishing robots.

5.2 Evolution of Robotic Surgery

The frst-ever documented robot-assisted surgery (RAS) was carried out in
1985, with a device known as PUMA 560 (Programmable Universal Machine
for Assembly or Programmable Universal Manipulation Arm). It was a single-
standard robotic arm, developed by Unimation, and was used to guide a
needle to the brain for biopsy, a procedure previously subjected to error due
to hand tremors. It was made possible by implementing computer tomog-
raphy (CT) for guidance. In 1998, PROBOT was developed by Imperial
College, London, and was used to perform transurethral prostate surgery as
the procedure required frequent repetitive cutting motions.
The next robot in line was ROBODOC, 1992, developed by Intuitive
Surgical Systems and IBM; it was used for creating a cavity in the femur
for hip replacement. The next in development was Automated Endoscopic
System for Optimal Positioning (AESOP) system, in 1990, by Computer
Motion, which was the frst system approved by the US Food and Drug
Administration (FDA) and was used for endoscopic surgical procedures.
ZEUS Robotic Surgical System was developed by the same company,
designed to assist in surgery. DaVinci Surgical System, the World’s most
used robotic surgical, was approved by the United States Food and Drug
Administration (US FDA) in 2000. Then, Intuitive Surgical bought Computer
Motion, thus discontinuing usage of the ZEUS Robotic Surgical System. Till
present, DaVinci has dominated the market of robotic surgery. Many other
robots have come into existence in the past decade and have achieved a
successful hold in the market in the feld of their usage.

5.3 Benefts of RAS

RAS has various advantages over traditional open surgery. The robots act as
extended and ergonomic hands of the surgeon that have better capabilities
and also cause less cutting and operation. The doctor can sit on a console or
 Robotics in the Medical Industry ◾ 61

use a hand-held device that can make the job much easier, hence enabling
him to perform a larger number of operations within the prescribed time,
with less stress and workload. RAS decreases the hospital stay time since it
avoids basic cutting (minimally invasive surgery); therefore, the patient can
get back to his physical strength in a matter of few days, in contrast to a
couple of weeks if performed traditionally. Since the recovery time is less,
the patient can get back to his normal life without having much to worry
about. RAS results in a lesser amount of blood loss and suturing. This pro-
cedure enables better visualization of the internal structure of the body,
through various interfaces in HD and 3D, which can act as eyes of the
surgeon. A few years ago, surgeons had to move the laparoscopic camera by
hand, and it only offered a vision in a single direction, but there have been
developments of a robotic camera that offer a wider feld of vision and thin-
ner arms to provide easier penetration. Gesture-controlled cameras are also
present that enable a surgeon to control the camera according to his com-
fort. Some of the software can take various MRI scans and combine them
to provide a single, refned image of our brain. Simulation software has
also been developed to provide lifelike operating experiences to the trainee
who is learning the use of robots. Knee and hip replacement robots offer
movement in 7 degrees of freedom and simulation of hand movements that
provides effcient placements of implants. The software also determines a
pre-operating plan that is to be followed by the robot, and if by any chance
a deviation occurs, it can be overridden manually.

5.4 Current Market Scenario

Diving into surgical robots, they can roughly be categorized into broad
classes, as shown in Figure 5.2.
Single Port Access (SPA) is known as the process in which a small incision
(2′–4′) is made on the body instead of a whole cut, thereby offering several
advantages:

Figure 5.2 Classifcation of technology used in surgical robots.

62 ◾ Machine Learning for Business Analytics

◾ Less amount of blood loss and lesser quantity of blood required during
the procedure
◾ Less post-operative stay of the patient in bed (2–3 days) which is much
higher in general open surgery (2 weeks to 1 month)
◾ Less stress on the surgeon therefore can perform a higher number of
operations
◾ Lesser room in making mistakes as human error is eliminated
◾ Minimal scarring as the operation is conducted using minimally inva-
sive surgery
◾ The amount of pain and discomfort is signifcantly less
◾ Better visualization of the internal body
◾ The ergonomic design of the robotic arm end.

SPA is used almost in all of the surgeries available to a human being—

general, spine and brain, cardiovascular, vascular catheterization, gynecol-
ogy, ocular, neurology, and urology. Figure 5.3 shows that 36% of the SPA
robots are currently in the pipeline.
Various companies have their commercialized robots and some of which
are still in the pipeline. Some of the most commonly used robots in the
industry are DaVinci (Intuitive Surgical), MAZOR X (MAZOR), Senhance
(TransEnterix), and Excelsius GPS (Globus Medical).

Figure 5.3 The breakup of Single Port Access market of surgical robots.

 Robotics in the Medical Industry ◾ 63

The number of robots currently in the market is larger in number than those
that are still in development. One of the most awaited and currently in devel-
opment is Verb Surgical, which is being headed by Alphabet (part of Google),
and its aim is to “democratize surgery.” It will have AI augmentation, which is
supposed to change the feld of surgery and is not launched completely yet.
Some technologies are required in all of the surgeries and are a must-
have for all the robots:

◾ Full wrist-like rotation of the robotic arms end to provide ergonomic

access
◾ Variable grip graspers so that the arm can be used for various functions
such as cutting, suturing, grasping
◾ Planning out preoperative 3D plans which are to be followed by the
robot thoroughly
◾ Mapping out a virtual boundary so that the robot doesn’t go around
and harm healthy tissues and organs
◾ Providing real-time imaging so that the surgeon can see where and how
it is operating through an endoscopic camera
◾ Providing haptic feedback in various degrees to create a sensation of
touch when operated by the robotic arm, which can help the surgeon
in identifying different tissues without physically feeling the tissue by
his/her hand.

These robotics technologies are still present only in well-developed coun-

tries, while each company is trying to make its product available for the
worldwide population. Looking at the following graph we can see the
growth in the usage of robots in performing total knee arthroplasty in
the United States from 2005 to 2014. There were a total of 60,60,901 cases
reported from 2005 to 2014 in which 2,73,922 were performed using com-
puter navigation and 24,084 were completed using robotic assistance. A
graphical representation of the data is shown in Figure 5.4 [7].
According to the fndings of a study by BCG, as of January 2016, the
global medical robotics industry clocked an annual revenue of $7.47 billion,
and this was destined to rise at a robust compounded annual rate of 15.4%
over the next fve years. Considering this, the global sales of medical robots
are expected to grow by $8.90 billion during 2022–2026 [8].
Even in the Middle East countries such as Saudi Arabia, there has been
considerable growth in robotic surgeries. The total number of RAS in the
Middle East is still low compared to Europe and the United States [9]. To
64 ◾ Machine Learning for Business Analytics

Figure 5.4 The growth in robots usage from 2005 to 2014.

Figure 5.5 The trend of the number of procedures performed globally in 2017.

date, there are over 44 hospitals with DaVinci robots installed in them and
ready to use for the general population [10].
Over the years, Intuitive Surgical has seen an exponential rise in usage
of their robots for various procedures such as urology and gynecology.
Figure 5.5 depicts the approximate number of RAS procedures that have
taken place all over the world.
Similarly, New Zealand was introduced to robotic surgery in 2017 by
Intuitive Surgical at Southern Cross [11]. Since then, the number of opera-
tions being held there is increasing signifcantly every year. Most DaVinci
robots are still in the United States, with the number being over 2,500, fol-
lowed by China and Australia.
South Korea is also not behind the other nations in using the robot to
conduct surgeries. A study shows the usage of DaVinci at a single institute
from 2005 to 2013 had steadily risen and seen the trend it will continue to
 Robotics in the Medical Industry ◾ 65

Figure 5.6 The growth in the robotic procedures performed in Yonsei University in

Korea.

grow in future also. During this period, robotic system surgery was used
for a total of 10,267 cases. Statistical analysis showed that 15 surgeons per-
formed 93.47% of all 10,000 surgeries. One of the surgeons performed more
than 2,500 surgeries, and two performed more than 1,400 each. Figure 5.6
shows the distribution of the mentioned surgeries over the years [12].
Figures 5.5 and 5.6 provide us with a general idea about how robotic sur-
geries are getting a grip on the industry.
This sector, which was dominated by a single company for the past few
decades, may witness fragmentation in the near future due to the development
of many surgical robots worldwide. Intuitive Surgical with its DaVinci series
of robots has dominated the global market and is a pioneer in this indus-
try. It continues to be a prominent developer and seller of the most widely
used robots in various surgeries and the most recommended ones by various
surgeons. To cite an example, MAKO is the robot that is used extensively in
total/partial knee/hip replacement surgeries. The market is getting disrupted
by many companies that are developing robots of their own, of which one of
the most awaited is Verb Surgical developed by Alphabet (part of Google).

5.5 Future Growth Estimates

Following are the estimates that the author has found through secondary
research or has quantifed according to previous data collected from various
statistics present:
66 ◾ Machine Learning for Business Analytics

◾ The worldwide market size of robotic devices is currently evaluated at

around $5 billion and is expected to multiply by fve times the amount
in the next couple of years [13].
◾ The market in India for medical robotics is going to reach as high as
Rs. 2,600 crore by the end of 2024, as stated by the company Vattikuti,
which plans to install a total of over 100 DaVinci robots by 2020.

And also train at least 500 surgeons on how to operate the robot to its full
potential [14].

◾ The system revenue of Intuitive Surgical, manufacturer, and distributor

of DaVinci robot would grow close to $1.5 billion by 2021 from $1.13
billion in 2018. It is presently having over 4,900 functional units all over
the world. Assuming constant growth, the number is expected to grow
to 6,000 by 2022 [15].
◾ A greater number of devices can be found in the United States, followed
by Europe and Asia. Every one in four hospitals in the United States has
a surgical robot. The manufacturers are planning to target highly popu-
lous countries like China and India for increasing their usage.
◾ The offcial notifcation issued by the Chinese Government had
expressed the requirement of 197 endoscopic surgical instrument
control systems, or surgical robots, by the end of 2020. That number
includes an additional 154 new systems of Intuitive Surgical [16].
◾ Intuitive Surgical is also aiming to set up their base in the military feld
also, as using 5G technology complex surgeries can take place without
the requirement of a surgeon to be on feld duty, but just controlling the
robot from his/her convenient place.

5.5.1 Reasons for the Lower Than Required

Presence of Surgical Robots
◾ The main reason for the under-penetration of robots in healthcare,
which has been stated by the experienced professionals and experts
in this feld regarding better exposure to robotic surgery, is the lack of
clinical evidence.
◾ Another reason is the reluctance of the ecosystem to change and adapt
with time. There is a lot of potential in this market, which people fail
to recognize for the fear of loss of jobs or fear of upskilling themselves.
We need some organizations that deal with gathering this evidence and
 Robotics in the Medical Industry ◾ 67

publishing it, to make people who are unaware aware of the various
benefts and advantages of RAS over traditional open surgery.
◾ Advertisement is another point to work upon. The medicos or the cus-
tomers must get to know about any product that comes to the market.
For this, the RAS device manufacturers need to promote the robotic sur-
geries to the doctors, who, in turn, need to promote them to the patients.
◾ The medicos have a lack of skill that is required for robotic surgeries.
If this skill can be imparted to the surgeons, the robotic surgery can
have the potential for inclusive healthcare, as the robots can be updated
with the latest software updates equally and the surgeon can operate
on the patient sitting at his/her desired location (Telesurgery).
◾ The robotic parts are costly to make, and they require replacement and
maintenance at a regular frequency, making it more costly unless the
suffcient scale is achieved.
◾ There is a loss of human touch of the doctor when the patients get
attended to by a robot, which gives RAS a lower edge over the manual
surgeries.

5.6 Limitations of Robotics

Till now, we have seen the benefts and the bright side of the application of
robotics in healthcare. However, like every other innovation, it has its disad-
vantages or hurdles that need to be surpassed to increase its usage.

◾ The most important of them is the learning curve required for the
operation of these robots. One just can’t come and start using them; a
surgeon undergoes various months of training to learn the abilities and
working of robotic assistance and to get the full potential out of it. The
patient stays at the hospital may be lessened, but the time required for
trainees to become profcient in the use of robots is much higher.
◾ The second advantage is the high cost over a small scale. Any technol-
ogy cannot become successful until and unless it is being used by the
common population, and looking at current numbers, it won’t catch
up with the people by costing nearly $2,000 per operation. Further,
the annual maintenance of the robot is also required, which adds up
to several hundred thousand dollars for the hospital. The price of each
machine goes up to “$1 million to $2.3 million, plus up to $170,000 per
year in maintenance, according to Intuitive Surgical” (Torres, 2011) [17].
68 ◾ Machine Learning for Business Analytics

◾ Thirdly, some robots require the patient to be in a certain position for

the robot to operate on them. If the patient lies in improper positions
during the whole operation, there may be cases in which the patient
cannot feel his/her legs or hands, called partial paralysis or positioning
injury.
◾ Fourthly, the usage of telesurgery requires a good data connection. A
fault that may occur in the absence of a strong internet connection may
be beyond the capabilities of the surgical team. Hence, all the risks
must be kept in mind, and countermeasures should be ready for each
failure mode that can occur.
◾ Fifthly, the issue with latency, which is the time required by the robot
to execute the command given by the robot, can create a problem. It
can create a problem for a surgeon to respond if any complications
occur.
◾ Sixthly, the RAS devices have relatively larger footprints on the space
and relatively more cumbersome robotic arms. This gives them a hands-
down in the crowded operating rooms, where it may be hard for both
the surgical team and the robot to ft into the operating room.
◾ The seventh disadvantage is the psychological fear of the population.
People are afraid to go under the knife and being operated on by a
robot. They feel more secure, if a human is performing the operation,
and psychological behavior plays a very important in the well-being of
a patient’s health post-operation.

5.7 Future Directions for Increasing the

Acceptance of Robotic Surgeries
There are still various voids that need to be flled and limitations that need
to be overcome through the innovations. Some of the future directions are
listed below:

◾ The operating system of RAS can be made more user-friendly and well
defned so that it can be learnt much faster, and augmentation with AI
to suggest further operations is also possible, which would save the
time of both patient and surgeon.
◾ The robotic arms need to have ergonomic features so to be able to oper-
ate through any angle without the hassle of the positioning of the patient.
This will prove to be benefcial for both the surgeon and the patient.
 Robotics in the Medical Industry ◾ 69

◾ The issue of latency can be solved by involving the technology of 5G,

which will also help in telesurgery as it involves the transfer of heavy
amounts of data in real time.
◾ The development of better laws and coverage by insurance in case of
mishaps during robotic surgery can help the patient in terms of money
and future operations.
◾ Currently, there are two main agencies that facilitate the verifcation and
commercialization of surgical robots, USFDA (followed in the United
States and some Asian countries) and EMA (followed in Europe). If
there can be a single, unifed agency for this job, the distribution and
commercialization can be much easier, and we can do away with the
need of applying in every administration before marketing.
◾ US Army and NASA scientists are taking a heavy interest in the develop-
ment of telesurgery to be able to attend to their soldiers and astronauts
in space so that the requirement of the physical presence of surgeons is
eliminated.
◾ Software being used in robots needs to be upgraded constantly to make
it user-friendly and provide a better user interface and lesser lag time in
between commands and execution.
◾ Various levels of haptic feedback can be developed to enable robotic
surgery in delicate operations such as optical surgeries.
◾ The development of a computer-based intelligent tutoring system for the
training of cognitive and procedural skills that are needed to complete
basic robotic suturing can be useful for novice surgeons [18, 19].
◾ Insurance companies need to be encouraged to also include robotic
surgery in the healthcare cover, which can also help in inducing the
sense of assurance that robotic surgery is safe and prominent future of
healthcare.

5.8 Conclusion
This chapter explains the current landscape of the robotic surgery industry.
The investment required in the development and operation is large, and
hence institutions that choose to acquire the RAS devices should plan out a
cost-effective method for their utilization. Currently, the geographical market
that has been captured by the medical robots is not diverse and can be seen
in heavy usage only in few top developed countries of the world, the United
States, of course, being at the top, followed by SEA, Europe, and Australia.
70 ◾ Machine Learning for Business Analytics

Although robotic surgery is currently being utilized in developed coun-

tries mostly, in the next decade, it will achieve a worldwide presence. It will
be replacing the extra physicians who need to be present in the operation
theatre, with the presence of just the surgeon and the patient, thereby less-
ening the workforce required. The benefts of minimally invasive surgery are
far greater than the risks it has; and since it can’t be performed by human
hand, robotic surgery will become a better option in the coming future. It
will bring signifcant changes to healthcare.
As conveyed in this study, the authors feel upbeat on the future of this
industry and move on to suggest that robots are the future of surgery
because robotic surgery has various advantages over general open surgery,
some of them being lesser post-operative trauma, lesser operative time,
precise robotics movements, a lower amount of blood required for opera-
tion, better and faster decision-making on the diagnosis and the course of
action, better and real-time documentation of the medical records of the
patients, a wider range of complicated surgeries that can be performed by
robots, and mass customization of the surgeries using AI and ML.
This study claims that robotics is the next big innovation in the health-
care sector as the market has been continuously growing for the last fve
years and will to grow exponentially in the coming years. The usage of
these robots is also growing as people are becoming aware of their benefts
and shunning the fear of being operated by a robot. Development in this
sector is only waiting to be explored as we can develop a new development/
news articles/published papers every other day, and the companies are more
ready than ever earlier to inculcate them and also bank upon them.
The chapter also puts forward how robotics can help in medical inclu-
sion, particularly in the nations that have very low per-capita availability
of skilled medical doctors. The authors express their belief that a time will
come when robotic surgeries will be much more economical than manual
surgeries. Finally, the study also cautions about the fip side of using robotics
in healthcare. One of the areas touched upon includes the loss of the per-
sonal touch in the treatment of a patient by a robot.

References
[1] Wilson, H. J., & Daugherty, P. R. (2018). Collaborative intelligence: Humans
and AI are joining forces. Harvard Business Review, July–Aug.
 Robotics in the Medical Industry ◾ 71

[2] Joel, M. (2012). Don’t be afraid of the robots. Harvard Business Review, Dec 18.
[3] Beane, M. (2019). Learning to work with intelligent machines. Harvard
Business Review, Sep–Oct.
[4] Nedelescu, L. (2015). We should want robots to take some jobs. Harvard
Business Review, June 5.
[5] Dhar, V. (2016). When to trust robots with decisions and when not to. Harvard
Business Review, May 17.
[6] US Food and Drug Administration Website. www.fda.gov/medical-devices/
surgerydevices/computer-assisted-surgical-systems
[7] Antonios, J. K., Korber, S., Sivasundaram, L., Mayfeld, C., Kang, H. P., Oakes,
D. A., & Heckmann, N. D. (2019). Trends in computer navigation and robotic
assistance for total knee arthroplasty in the United States: An analysis of
patient and hospital factors. Arthroplasty Today, 5(1), 88–95
[8] Chengdu Tianfu Software Park. (2018). Borns: Breakthrough and prospect of
AI-powered minimally invasive surgery. www.prnewswire.com/news-releases/
borns-breakthroughand-prospect-of-ai-powered-minimally-invasive-surgery-
300675230.html
[9] Rabah, D. M., & Al-Abdin, O. Z. (2012). The development of robotic surgery in
the Middle East. Arab Journal of Urology, 10(1), 10–16. Doi:10.1016/j.aju.2011.
12.001
[10] Azhar, R. A., Elkoushy, M. A., & Aldousari, S. (2019). Robot-assisted urologi-
cal surgery in the Middle East: Where are we and how far can we go? Arab
Journal of Urology, 17(2), 106–113. www.ncbi.nlm.nih.gov/pmc/articles/
PMC6600062/
[11] Ford, E. (2016). Surgeons perform frst robotic surgery in New Zealand
to treat oral cancer. www.stuff.co.nz/national/health/86913698/
surgeons-perform-frst-robotic-surgery-in-newzealand-to-treat-oral-cancer
[12] Koh, D. H., Jang, W. S., Park, J. W., Ham, W. S., Han, W. K., Rha, K. H., & Choi,
Y. D. (2018). Effcacy and safety of robotic procedures performed using the da
Vinci Robotic Surgical System at a Single Institute in Korea: Experience with
10000 cases. Yonsei Medical Journal, 59(8), 975–981. https://doi.org/10.3349/
ymj.2018.59.8.975
[13] Jinoy Jose, P. (2018). Da-Vinci’s 5 million surgeries. A blog article. www.the-
hindubusinessline.com/opinion/columns/the-cheat-sheet/da-vincis-5-million-
surgeries/article24873712.ece
[14] Trefs Team. (2019). How much can intuitive surgicals system revenue grow
over the next three years. A blog article. www.forbes.com/sites/greatspecula-
tions/2019/07/09/how-much-canintuitive-surgicals-system-revenue-grow-over-
the-next-three-years/#67d975bf4418
[15] Hartford, J. (2015). Intuitive surgical makes the case for robotic surgery.
Medical Device and Diagnostic Industry, 37(7).
[16] Dansford, F. (2018). Intuitive surgical shares rise on expanded surgical robot
quota in China. A blog article on Mass Device. www.massdevice.com/
intuitive-surgical-shares-rise-onexpanded-surgical-robot-quota-in-china/
72 ◾ Machine Learning for Business Analytics

[17] Toress. (2011). Political implications and robotic surgery. A blog article. www.
bartleby.com/essay/Political-Implications-And-Robotic-Surgery-PKLNH6FMZRPA
[18] Bertelsen, A., Melo, J., Sánchez, E., & Borro, D. (2013). A review of surgical
robots for spinal interventions. International Journal of Medical Robotics, 9(4),
407–422. Doi:10.1002/rcs.1469
[19] Julian, D., & Smith, R. (2019). Developing an intelligent tutoring system
for robotic-assisted surgery instruction. International Journal of Medical
Robotics and Computer Assisted Surgery. https://onlinelibrary.wiley.com/doi/
abs/10.1002/rcs.2037
Chapter 6

Integration of AI in
Insurance and Healthcare:
What Does It Mean?
A. Kannan, B. Justus Rabi, and M. Anand

Contents
6.1 Introduction ..............................................................................................74
6.1.1 The Industry’s Major Issues Are Divided
into Six Categories........................................................................74
6.2 Review of the Literature...........................................................................75
6.2.1 Artifcial Intelligence ....................................................................75
6.2.2 Implementation of AI in Business ...............................................75
6.2.2.1 Insurance Business Models ...........................................77
6.2.3 AI in Customer Experience..........................................................77
6.2.3.1 Natural Language Processing ........................................77
6.2.3.2 Voice Analytics...............................................................77
6.2.3.3 Machine Learning ..........................................................77
6.2.4 Artifcial Intelligence in Claims....................................................77
6.2.5 Medical Imaging and Diagnostics ...............................................78
6.2.6 Insurance for Self-Driving Cars....................................................78
6.2.7 Person to Person Insurance .........................................................78
6.2.8 Insurance for Smart Homes .........................................................79
6.2.9 Health Insurance ..........................................................................79
6.2.10 Insurance for Aerial Assessors .....................................................79
6.2.11 Car Insurance ...............................................................................80

DOI: 10.4324/9781003206316-6 73
74 ◾ Machine Learning for Business Analytics

6.3 Research Methodology.............................................................................80

6.3.1 We Took SERVQUAL Model to Conduct a Methodology
Which Includes Five Dimensions as Follows ..............................81
6.4 Discussion of the Findings ......................................................................82
6.4.1 Recognition of the Face ...............................................................83
6.5 Conclusion................................................................................................84
References.........................................................................................................85

6.1 Introduction
Insurance often historically had a poor level of consumer involvement.
Insurers are the industry in which consumers engage the least, according
to the study. As insurance and healthcare is not completely digitized, the
healthcare and insurance providers did not get any opportunity to actually
understand what the customer needs and offer tailor-made solutions [1].

6.1.1 The Industry’s Major Issues Are Divided

into Six Categories

Right Advice: Offering a best bundle of products that meet consumer

needs. The cost of missing chance: walking in the streets to meet
potential customers early without delay.
Time-consuming: Provides quickest claim assistance to loyal customers.
Cost: Increased cost of claims put companies on the losing side in order
for them to make a marginal proft [1]. Disruptive models can arise as a
result of the IoT, Big Data, and the access to view larger environments
than previously possible. These may have a signifcant effect on the
insurance industry across the value chain, altering risk assessment and
evaluation as well as consumer engagement models [2].

When insurers invest in digitization, we see advantages for tech giants. We

believe the biggest obstacles will be in customer channels and emerging
data sources (Internet of Things and Big Data). Investment in core infrastruc-
ture would also be needed to allow digitization and cost reduction. Vendors
who sell in such areas should see a great deal [2].
The study looks at how technology is starting to reorganize the insurance
environment, as well as the industry’s future threats and opportunities [3].
 AI in Insurance and Healthcare ◾ 75

6.2 Review of the Literature

The research is categorized into two sections: the power of artifcial intel-
ligence (AI), which has emerged as a tech giant, and the obstacles that AI
encounters in being a catalyst to the insurance value chain [4].
Diffculties to AI in healthcare emerge/root from a variety of origins tak-
ing into consideration the insurance companies, regulatory authorities, and
the policyholders. Diffculty in making decisions centers the diffculties of
AI. Increasing volume of data is another challenge that has been observed
that AI feels diffcult to deal with [4].
AI is a computational course that is system-assisted and that aims to cre-
ate intelligent automated systems. In terms of intellect, AI is categorized as
fragile, robust, and bright AI [1].
The report explains the historical problems that have driven various
forms of actuarial methods in the literature review sections before going on
to review the machine learning (ML) and AI strategies that have been used
by analysts in doing such analysis [5].

6.2.1 Artifcial Intelligence

The intellect levels of AI are categorized into fragile, robust, and brilliant AI:

Fragile AI, also known as Artifcial Narrow Intelligence (ANI), is a form

of AI that focuses on a single task.
Robust AI, also known as Artifcial General Intelligence (AGI), is a form
of AI that mimics human intelligence.
Brilliant AI (ASI5) is an AI capable of innovative and scientifc reasoning
that surpasses human intelligence [1].

How AI is very smart in global is shown in Figure 6.1.

AI interacts with the atmosphere in the way mentioned in Figure 6.2. AI
gathers input sent by the atmosphere and makes analysis based on the input
gathered and preceding knowledge. It accepts text in a visual and picture
format and analyzes the date to offer powerful solutions [1].

6.2.2 Implementation of AI in Business

For some of our interviewees, we addressed the effects of AI deployment in
the workplace. Rearden (2019) said in a personal interview that AI will be
76 ◾ Machine Learning for Business Analytics

Figure 6.1 How smart is artifcial intelligence?

Figure 6.2 Sense-think-act process followed by an intelligent agent.

employed in the future to search the universe of knowledge and suggest the
most optimal and socially conscious techniques that create value and tactics
to assist businesses in identifying and leveraging strategic plans.
The adoption of AI, according to Rearden, could result in the following
outcomes:

◾ Value generated in society improves brand recognition and

attractiveness.
◾ The value that has been created for workers decreases morale, increases
effciency, and strengthens the company. All of these factors work
together and enhance the consumer experience.
◾ Long-term longevity and higher income.
◾ Increased chance of successful plan execution and lower likelihood of
failure [5].
 AI in Insurance and Healthcare ◾ 77

6.2.2.1 Insurance Business Models

Insurance business models (BMs) can be explained in a variety of ways,
including by plotting value chains. Since we live in a data-driven environ-
ment, another way to understand a BM is to look at data, its origins, and
how it is used. Another factor that affects the prevalence of BMs are plat-
forms such as Google and Alibaba.
There are standard insurers’ business managers, who perform a variety
of operations on websites such as the billing methods and storage of
information [4].

6.2.3 AI in Customer Experience

6.2.3.1 Natural Language Processing
Data mining, subject modeling, and emotion analysis of unstructured web-
based social engagement data are all used in natural language processing.

6.2.3.2 Voice Analytics

Call center audio recordings are used to explain when people call and how
they feel.

6.2.3.3 Machine Learning

Analysis of the decision tree, learning developed by Bayesian and social sci-
ences are examples of ML techniques that can be used to determine behav-
ior from information [6].

6.2.4 Artifcial Intelligence in Claims

Soft robotics: Process mining tools are used to fnd constraints and
increase effciencies and compliance with traditional claims
procedures.
Chart Analysis: Graph or social network analysis is used to detect fraud
trends that are present in claims.
Machine Learning: Deep learning methods are used to automatically
categorize the sort of damage to cars involved in collisions in order to
assess repair costs. To construct claims predictive models, use decision
trees, support vehicle machines, and networks of Bayesian.
78 ◾ Machine Learning for Business Analytics

Internet of Things: Data of IoT is used to create operational knowledge

on the occurrence and nature of incidents to offset danger and elimi-
nate damages [6].

6.2.5 Medical Imaging and Diagnostics

The Food and Drug Administration certifed an AI Program in April 2018 that
tests patients for diabetic retinopathy without requiring expert advice from a
specialist IDx-DR, which is a software that accurately classifes patients with
retinopathy diabetics 84.7% of the time, shown in Figure 6.3 [7].

6.2.6 Insurance for Self-Driving Cars

Certain collisions, such as those caused by environmental and human
causes, will be impossible to prevent even with self-driving vehicles. In six
years of texting, self-riding automobiles [8] invented by Google crashed
almost and nearly for 11 times as of now.
Self-driving vehicles can pose new risks in a variety of ways, from mal-
function of the self-driving process and hacking [9].

6.2.7 Person to Person Insurance

◾ A person-to-person insurance is one of the insurance models (shown in
Figure 6.4), which can be made possible by social media, which allows
insurance providers to form online networks to exchange risks.

Figure 6.3 Diabetic retinopathy using AI.

 AI in Insurance and Healthcare ◾ 79

Figure 6.4 Insurance model.

◾ To share the risk, customers can enter or create their own online social
networks.
◾ Members of the network contribute a part of their premium to a fund
for shared beneft.
◾ The balance of the premium is paid to the insurer by the members [9].

6.2.8 Insurance for Smart Homes

Premium discounts are available to State Farm policyholders who install
ADT pulse, a home protection device that helps to minimize leakage from
a damaged tube, therefore avoiding huge expenditure by getting real-time
signals from sensor devices that identify problems of fxtures, among other
issues [3].

6.2.9 Health Insurance

Google is collaborating with Novartis, a Swiss-based global healthcare pro-
vider, for developing a canny wear lenses that try to correct visibility as well
as control glucose volumes, which could be life-saving for diabetics [3].

6.2.10 Insurance for Aerial Assessors

Drones, otherwise known as lightweight air-borne vehicles, are arousing the curi-
osity of insurer frms because they offer effcient methods for insurers to enhance
business functionalities such as compensation processing and endowing.
80 ◾ Machine Learning for Business Analytics

6.2.11 Car Insurance

Car insurance is yet another category where connected device can have an
effect, as telematics can relay useful data for determining a person’s risk level.
In place of depending solely on details such as life time, sex, the type of
hooptie someone drives, and old history of accidents, insurance companies
can get access to updated information on their policy-holders’ riding behav-
ior, such as whether they make sharp turns or sudden stops, determining
the speed with which they travel in a car, how often, at what period of a
day, and region.

6.3 Research Methodology

Since the future of AI is far beyond what has been observed for now and
can build a brand new technological world of healthcare, in locations that
reject the prevailing hypothesis, an explanatory approach was chosen. The
study examined case lets of healthcare companies and insurers that used AI
and performed discussions with veterans [4].
Educational journals, practitioner-focused sources, websites of com-
panies, website of authorities who support the supply chain of insurance
companies, parent companies that are owned by insurers, and employees
working in an insurance company were among the data sources [4].
Theory was developed using case study research. This inductive method
was applied to a total of 20 cases. It was not a random selection; it was
focused on how potentially benefcial they are. The cases were selected
because they covered a wide range of topics such as incumbents and dis-
ruptors, and various geographic areas with various degrees of technological
emphasis [4].
Case lets were assessed using the concocts mentioned in the review of
literature. The concocts are optimized and validated in a repeated cycle
process. Case-by-case review was carried out using the frameworks. One of
the writers with an experience in healthcare fnancials and insurance orga-
nizations acted as a “local sinner’s lawyer” to boost the study’s reliability [4].
Finally, group discussions inviting 12 veterans from major InsurTech
companies were conducted. Many of the experts had prior experience with
AI implementation in the insurance industry [4].
Their responsibilities include senior executives in insurers’ IT depart-
ments and insurance review portals, as well as insurance technologies
 AI in Insurance and Healthcare ◾ 81

providers and consultants in the feld of technology implementation.

Overall, they had past experience of 5 years in this industry [4].
The analysis of the current literature on the need for ML and AI in actuar-
ial practice was our primary research approach. To fnd applicable literature
for our study, we created the following structure [5].
Journals not described previously, such as the Actuarial Journal of North
American and the Actuarial Journal of South African, were also mentioned.
However, there was no related material in such actuarial papers [5].
This approach has not always resulted in an adequate volume of litera-
ture. As a result, we augmented our primary study with:

a. Interviews with people working in the area of AI.

b. Extending the reach of our study to include data from Kaggle [5].

As a basis for picking the sample, this study uses non-probability sampling
techniques. Surveys emailed 150 copies of the questionnaire:

1. We have sent mails and fax to few selected and known clients, director
and manager of the project, the insurance companies, policy brokers,
and advisors of the claim.
2. To Chinese building management scholars [10].

The research paper mainly targets clients, contractors, and insurers of the
Chinese economy. The total responses we received sum up to 41. They
cover a wide range of regional areas of activity, as well as a wide range of
industry experience, company size, foreign market experience, credentials,
and contractor styles [10].
This study’s research approach is divided into two sections. The evaluation
criteria are mentioned in the frst step based on the SERVQUAL model’s fve
dimensions. According to the Central Insurance of Iran’s 2012 yearbook, 13
insurance companies were chosen for evaluation [11].

6.3.1 We Took SERVQUAL Model to Conduct a Methodology

Which Includes Five Dimensions as Follows
1. Materiality: Physical factors, new facilities and equipment, personal
presence, and the organization’s physical environment.
2. Authenticity: The ability to provide services quickly and consistently.
82 ◾ Machine Learning for Business Analytics

3. Assurance: Workers knowledge and ability to establish mutual conf-

dence between consumers and service providers.
4. Responsiveness: Willingness to provide timely services and to assist
customers.
5. Empathy: When providing services, service providers pay close atten-
tion to the needs of their customers (availableness and comprehending
of clients).

The SERVQUAL allows for the assessment of service quality from the view-
point of the consumer, as well as the tracking of customer preferences and
perceptions over time, as well as the differences among themselves.

6.4 Discussion of the Findings

In fndings, we’ll look at how effective the XGBoost algorithm is at predict-
ing risky clients and potential claims. The effciency of physical and digital
ML classifcation algorithms to recognize and identify various forms of fraud
is then assessed. We are presenting few of the outcomes on implementation
of the framework of the block chain [12] in Tables 6.1–6.3.
We used various regression ML algorithms to solve the potential claim
since it is a regression problem.
The type of data is considered to be Gaussian in the Naive Bayes algorithm.
In addition, for the nearest neighbor, we took the count of neighbors to
the same count of classes identifed as fraud. We applied equal weighted

Table 6.1 Client Risk Rate (Performance Table).

Classifer Accuracy (%) Precision Recall F1-Score

Decision tree 74.44 0.6473 0.5953 0.6005
SVM 73.21 0.6696 0.5652 0.4841
Nearest neighbor 73.80 0.6696 0.5256 0.4841
XGBoost 76.81 0.6828 0.6295 0.6392
 AI in Insurance and Healthcare ◾ 83

Table 6.2 Fraud Detection (Performance Table).

Classifer Accuracy (%) Precision Recall F1-Score Training Time (ms)

Decision tree 92.99 0.870 0.929 0.892 471
Naive Bayes 52.06 0.373 0.520 0.425 155
Nearest neighbor 42.70 0.223 0.427 0.255 1254
XGBoost 99.25 0.9928 0.992 0.9926 995

Table 6.3 Confusion Matrix

Predicted Observed True False

Positive True positive (TP) False positive (FP)
Negative False negative (FN) True negative(TN)

uniform distribution. We took the same data for all the classifers and segre-
gated the train and test data in proportion of 70:30.
To solve the result of the classifcation, we considered the confusion
matrix to calculate the following scores [12].
The attitude of contractors, according to 34.29% of respondents, is an
important metric driving the growth of construction in terms of manage-
ment of risk appetite. In China, culture is found to have a signifcant impact
behind the growth of healthcare management. Chinese people used to be
known for their conservatism.
Managing risk capabilities is essential for operations and project manage-
ment according to 86% of those polled. The people having practice continue
catching hold of danger by its cost and nothing to do with attitude of the
old system [10] as shown in Figure 6.5.

6.4.1 Recognition of the Face

Face detection is becoming more popular, from unlocking mobiles to board-
ing planes. As far as facial recognition is concerned, China’s unabashed drive
for surveillance, along with its AI aspirations, has dominated the news as
shown in Figure 6.6.
84 ◾ Machine Learning for Business Analytics

Figure 6.5 Importance versus frequency.

Figure 6.6 China’s facial recognition trends up in news mentions.

6.5 Conclusion
Integrating AI in insurance results in improved loyalty of the customer, frm
earnings and reduction of the fraud, good time management, and orga-
nizational uncertainties. AI use concepts backed by real-life incidents of
corporates, demonstrating the vast potential of the insurance sector. New
 AI in Insurance and Healthcare ◾ 85

approaches to monitor, manage, and price risk, interaction with consumers,

cost cutting, increase performance, expansion of the insurance industry all
being enabled by emerging technologies and developments. Back-testing
parametric mortality models can be done using ML techniques, especially
the RT enhancing system. These methods allow us to identify the faws in
such models using real-world evidence.
Radiation therapy enhancing can be used to predict cause-of-death mor-
tality rates from actual evidence, according to research on cause-of-death
mortality in a Poisson model setting. This method makes it possible to spot
trends in these odds over time.
This study uncovered information at the business and corporation levels.
The focus groups spoke about transformative ability of the AI, which is sup-
ported by other fndings. Using ML analytics in insurance to develop mar-
keting campaigns, grow the company, increase revenue, and cut costs. The
consistency of which statements are predicted will have a direct bearing on
the actual economy.

References
[1] Kumar, N., Srivastava, J. D., & Bisht, H. (2019). Artifcial intelligence in insur-
ance sector. Journal of the Gujarat Research Society, 21(7), 79–91.
[2] Hocking, J., Wood, A., Dally, N., Pan, K., Lin, B., Ban, H., . . . Lee, S. (2014).
Insurance and technology: Evolution and revolution in a digital world. Blue
Paper, Morgan Stanley Research.
[3] Institute of International Finance. (2015). Innovation in insurance: How tech-
nology is changing the industry. European Banking Authority.
[4] Zarifs, A., Holland, C. P., & Milne, A. (2019). Evaluating the impact of AI on
insurance: The four emerging AI-and data-driven business models. Emerald
Open Research, 1(15), 15.
[5] Literature review: Artifcial intelligence and its use in actuarial work. (2021).
Retrieved April 11, 2021, from www.soa.org/globalassets/assets/fles/resources/
res earch-report/2019/ai-actuarial-work.pdf
[6] Yoder, J. (2016). AI in insurance: Hype or reality? [online] Digital Insurer.
Retrieved April 11, 2021, from www.the-digital-insurer.com/wp-content/
uploads/2016/06/716-pwc-top-issues-artifcial-intelligence.pdf
[7] Pfeifer, R., & Iida, F. (2004). Embodied artifcial intelligence: Trends and
challenges. In Embodied artifcial intelligence (pp. 1–26). Berlin, Heidelberg:
Springer.
[8] Hanafy, M., & Ming, R. (2021). Machine learning approaches for auto insurance
big data. Risks, 9(2), 42.
86 ◾ Machine Learning for Business Analytics

[9] Hurley, R., Evans, P., & Menon, A. (2015). Insurance disrupted: General insur-
ance in a connected world. London: The Creative Studio, Deloitte.
[10] Liu, J., Li, B., Lin, B., & Nguyen, V. (2007). Key issues and challenges of
risk management and insurance in China’s construction industry. Industrial
Management & Data Systems.107 (3),382-396.
[11] Saeedpoor, M., Vafadarnikjoo, A., Mobin, M., & Rastegari, A. (2015). A servqual
model approach integrated with fuzzy AHP and fuzzy topsis methodolo-
gies to rank life insurance frms. In Proceedings of the International Annual
Conference of the American Society for Engineering Management (p. 1).
American Society for Engineering.
[12] Dhieb, N., Ghazzai, H., Besbes, H., & Massoud, Y. (2020). A secure ai-driven
architecture for automated insurance systems: Fraud detection and risk mea-
surement. IEEE Access, 8, 58546–58558.
Chapter 7

Artifcial Intelligence in
Agriculture: A Review
Harshitha Sirineni, Thakur Santosh, and S. Deepajothi

Contents
7.1 Introduction ..............................................................................................88
7.1.1 What Is Artifcial Intelligence ......................................................88
7.1.2 Evolution of AI .............................................................................88
7.2 Role of AI in Agriculture..........................................................................89
7.2.1 Agriculture in India ......................................................................89
7.3 Applications and Techniques of AI Used in Agriculture ........................90
7.3.1 Crop and Soil Monitoring ............................................................90
7.3.2 Predictive Agriculture Analytics...................................................90
7.3.3 Agri Supply Chain ........................................................................91
7.3.4 Drones ..........................................................................................91
7.3.5 AI Agriculture Bots ......................................................................91
7.3.6 Precision Farming .........................................................................92
7.4 Future of AI in Agriculture ......................................................................92
7.5 Challenges in AI Adoption in Agriculture ...............................................93
7.6 Limitations ................................................................................................94
7.7 Conclusion ................................................................................................94
References .........................................................................................................95

DOI: 10.4324/9781003206316-7 87
88 ◾ Machine Learning for Business Analytics

7.1 Introduction
There will be a beginning for every miracle in this universe, where the
beginning must link with something before it happens. It may be beyond
our imagination, like say “what if machine thinking like a human.” This
thought in past we called as artifcial intelligence (AI) in present. So, let us
see the journey of AI from the 19th century to the 21st century.

7.1.1 What Is Artifcial Intelligence

An original aspect of AI is reenacting a higher capacity of human brain by
programming it to utilize natural language, masterminding speculative neu-
rons where the goal is to shape ideas and furthermore act like people, an
approach to decide and quantify issue intricacy, personal development, and
arbitrariness and innovativeness. AI isn’t a person versus machine saga; it’s
actually man with machine synergy [1].

7.1.2 Evolution of AI
AI is a well-known concept; its roots go back to the previous time in the
ancient Greek era. It was less than a century ago. From then, AI took off
high-tech revolution and went from fction to a very possible reality. In 1884,
Charles Babbage worked on an automated machine that will perform intel-
ligent behavior, but he decided that he would not be able to demonstrate the
progress of the machine, which he wanted to do, so he dropped his work in
the middle. A few years later in 1950, Alan Turing, known as WWII code-
breaker and a British mathematician, was universally accepted as being the
frst person to come up with an idea that a machine can think. He intro-
duced the Turing test, which is used till today, to gauge the thinking abil-
ity of a machine in comparison to human. Although his theory at the time
was ridiculed, later a proposal was developed in which the term “Artifcial
Intelligence” was widely known.
In the summer of 1956, Dartmouth College Professor “John McCarthy” coined
the term “Artifcial Intelligence” and is widely known as the “father of AI.”
In 1959, Marvin Minsky, an American intellectual scientist, light-
ened up the AI lamp amid the darkness of problems; he co-founded the
Massachusetts Institute of Technology’s AI laboratory. As a result, work on
AI continued by Marvin Minsky—one of the leading thinkers through the
1960s and 1970s in the feld. He even specifed in his book Stormed Search
 Artifcial Intelligence in Agriculture: A Review ◾ 89

for Artifcial Intelligence that the problem of AI modeling within a genera-

tion will be solved and suggested Stanley Kubrick, a flm-maker in 2001:
Space Odyssey was released in 1968, which gained worldwide recognition
and became one of the best AI presentations.
In the early 1980s, personal computers came into use, which grabbed
everybody’s attention and created more interest in machines that think like
a human. It took a couple of decades for people to realize the true impor-
tance of AI. Nowadays, people are interacting with AI daily in both physically
and virtually. For example, smart cars and IoT devices, and voice recognition
such as Apple’s Siri, Google Assistant, Samsung’s Bixby, Amazon’s Alexa, and
Microsoft’s Cortana. Here the common applications are machine learning (ML)
and neural networks. Other applications are deep learning and unsupervised
learning, which are the future of AI in the creation of high-level thoughts [2].

7.2 Role of AI in Agriculture

The agriculture sector frmly and openly grabbed latest technologies such
as AI to improve their productivity and to change overall outcome. AI solu-
tions support overcoming traditional challenges across all sectors in the
same way that AI in agriculture improves the fow of information, lowers
transaction costs, and speeds communication [3]. These have proven to be
the major drivers of economic growth. The agricultural sector discharges
have decreased by 20% due to AI changing the method of our food produc-
tion. Any unexpected natural conditions can be controlled and managed by
adapting AI technology.
In today’s world, with AI-enabled methods to boot agricultural produc-
tion effciency, AI is keeping an eye on startups in agricultural sector. To
reduce the rate of unfavorable outcomes, businesses in agriculture are pro-
cessing agricultural data with the help of AI. Furthermore, AI has improved
the applications in agriculture to identify diseases or climate changes faster
and respond intelligently, which is currently being implemented to improve
agricultural production effciency [4].

7.2.1 Agriculture in India

The origin of our culture is in agriculture. India’s agriculture is composed
of many different crops. The agriculture situation is ongoing ineffciency
of harvesting, transport, and storage of government-subsidized crops, and
90 ◾ Machine Learning for Business Analytics

environmental disturbances affected one-ffth of total agriculture. Today, we

are in the fourth industry revolution of agriculture. All four industry revolu-
tions are described below:

1. The frst upheaval depends on mechanical creation gear driven by

water and steam power.
2. The second depends on large-scale manufacturing empowered by the
division of work and the utilization of electrical energy.
3. The third depends on the utilization of gadgets and IT to additional
computerized creation.
4. The fourth is predicated on the utilization of cyber-physical systems.

Compared to other industries, the agricultural sector has been slow to imple-
ment, and to empower the fourth industrial revolution, we must adopt vari-
eties of technologies [4].

7.3 Applications and Techniques of AI Used in Agriculture

7.3.1 Crop and Soil Monitoring
AI is used to increase the value of each acre and deep learning for image
analysis. It claims to be a self-evolving system with intelligence that gives
farming solutions that are future-ready to the agriculture sector. This func-
tioning is carried out by using different IoT-based technology and sensors
to monitor plant and health of the soil. Deep learning helps us to identify
objects, fora, fauna, and tags them in the image. With this technology, agri-
culture product grading is done by automated quality analysis from the image,
and this alerts on crop infestation and suggests prevention measures [5].

7.3.2 Predictive Agriculture Analytics

AI-based sowing application uses various tools to intimate farmers
when to sow seeds, which is the biggest challenge. An advanced feature
added to this application is that it indicates risk of pest-based attacks
based on climatic conditions and the stage of crop. This information will
be sent to farmers in the form of automated text messages or through
voice calls [5].
 Artifcial Intelligence in Agriculture: A Review ◾ 91

7.3.3 Agri Supply Chain

Agri supply chain [5] is for analyzing real-time data on data streaming
that comes from various sources throughout the country. Agri online
market place suggests affordable prices for both consumers and produc-
ers with the data-driven technology. To ensure the working of Agri sup-
ply chain application is fast and effcient, various technology stages were
introduced:

◾ Transition discovery means it analyzes real-time data on various data

streams in addition to data obtained by the crowd. This helps us to
discover automatic transaction to make high margin for the beneft of
producers and consumers at the same time.
◾ Quality maintenance refers to automatic sorting and grading on
AI-based and computer vision for trading good quality across country.
◾ Credit risk management handles the credit default problem that rises in
supply chain to reduce operation risk rate.
◾ Agri mapping analyzes images that are captured by satellite, gathers
information with the help of deep learning, and obtains maps in real
time at high resolution.

7.3.4 Drones
Drones are capable of in-depth feld analysis, long-distance spraying, and
effcient crop monitoring. All these works are done by sensors that capture
images and videos; identify nutrient defciency, pest damage, fertilizer needs,
water quality, and chemical composition of surface; and track the surface
temperature and lidar sensor shows farm in 3D model. Compared to most of
the farmer’s equipment, drones are affordable [6].

7.3.5 AI Agriculture Bots

Agriculture bots are AI-enabled and used for helping farmers in order to
protect the crop from weeds in the most effcient manner. This can be done
by monitoring weeds with the help of computer vision and spraying them.
By using this application, we can harvest high volume of crops at a faster
rate than human labor [7].
92 ◾ Machine Learning for Business Analytics

7.3.6 Precision Farming

The goals of precision farming are proftability, effciency, and manageabil-
ity. There are key innovations that empower accuracy in cultivation: high
exactness situating framework, mechanized guiding framework, geo plan-
ning, sensor and far off detecting, incorporated electronic correspondence,
and variable rate innovation. These technologies provide advice about crop
rotation, water management, nutrient management, optimum planting and
harvesting time, pest attacks, and many. Precision farming can be summed
as a “Right place, Right Time, Right Product” [6].

7.4 Future of AI in Agriculture

Assuming you need to take care of the world in the future, we’ll need to
deliver a same amount of food in the next 40 years as we did in the previ-
ous 8,000 years, which shows a hint of the food framework’s confict. With
the development in populace, change in utilization conduct, and environ-
ment emergency, how would we get our food creation?
The genuine mystery is the manageable creation. It ought to be with less
data sources, less composts, less pesticide; and less water everything should
be manageable. Else we will annihilate our plants. The security of the food
framework is one of the world’s most squeezing diffculties. It all relies upon
how we tackle it; if everybody on earth is on diet of the normal American,
then that would require all tenable land to be utilized for horticulture, and
we are as yet being 38% short. Also, if that is correct now, then what do we
do when there are two billion additional individuals?
All things considered, the key is surprisingly energizing, and that is pro-
fciency. Essentially, how would we deliver signifcantly more on the land
that we are as of now utilizing? At the point when we need to go for feasible
creation, everyone should be a part of the modifed framework and embrace
development to achieve our goal of making profciency in your work that is
unrivaled anywhere else on the world.
That strategy is vertical cultivation; those are distribution centers with
piles of tank-farming systems to improve the growth of verdant crop. They
are being developed in urban communities everywhere in reality where new
produce and land are scant. The critical deterrent here is the expense of
energy and the cost utilizing a ton of it takes on the climate. The potential
gain is that counterfeit lights and environment-controlled structures permit
 Artifcial Intelligence in Agriculture: A Review ◾ 93

developing day and night, all year, creating an altogether better return for
every square feet than an external farm. For now, however, just costly, ver-
dant greens or spices have shown high yield in the vertical framework. In
addition, the jury is certainly still at a loss as to whether this is frequently
really harmless to the ecosystem method; potential arrangement is made to
use blue and red light frequency of photosynthesis development and a high
growth rate. Another development in the home is open farming work, which
means to make “inventory of environments” so temperature and moistness
can be set to reproduce the ideal conditions for developing harvests that
may ordinarily come from wherever in the planet, locally all things consid-
ered. This is frequently a push to handle the food miles issue [8].
When the delivered are dispatched all throughout the planet, it makes point-
less CO2 emanations. What’s more, “Our Future is about Accepting Challenges.”

7.5 Challenges in AI Adoption in Agriculture

The Agri input companies face many challenges when they try to manage
a demonstration farm; these farms are known as Demo Farms. Nowadays,
Agri input companies utilize these farms in various regions to showcase new
seed varieties or agro chemicals or to demonstrate new Agri culture tech-
niques in parallel to the existing traditional ones. Let us see some challenges
that Agri input companies face: collection of data across regions is a slow
and long process, and the manual work gives more chances to errors such
as loss of information. Because of that only pen and paper spread sheets or
elementary online tools are being used for data acquisition and processing.
So, what can be done to boost farmer loyalty with dozens of competitive
products battling for farmer attention. Agri business companies are also fac-
ing diffculties in retaining the loyalty of farmers. Designing effective farmer
engagement programs is also becoming very diffcult in the absence of reli-
able farm-level intelligence. The lack of real-time data to understand the fac-
tors that impact crop growth such as climate change and resulting weather
conditions, pests, and diseases, availability of soil nutrients, and water stress
level affects agricultural productivity. Is there any comprehensive solution for
these problems?
Yes, Agri input companies give accurate ground data, schedule daily
task, manage farm activities, and much more like looking for weather- and
satellite-based advisory to stay on top changing conditions and help farm-
ers to adopt a smart package of practices that reduces production costs and
94 ◾ Machine Learning for Business Analytics

improves effciency. For the management team, they provide access to cus-
tomize farm reports’ ability to provide real-time guidance on feld requests
and measure staff performances. The entire crop production cycle monitors
health and accurate yield estimations to make sure that farmers are ready for
sustainable and climate-resilient farming [7].
With all these challenges if we add up the most powerful challenge, that
makes scale productivity, increases effciency, and strengthens sustainability.
That challenge is image recognition to identify pathogens pest infestation in
crops. If we can identify them early enough, we can spray fewer chemicals
that lead to less crop distraction, leading to bigger crop yields that could
mean more food. To make this challenge happening, we need to build a
better database. These are the challenges AI faces in the feld of agriculture.
“Challenging for Good, we see a Better Future.”

7.6 Limitations
It costs a lot of money to make or buy the technology. Indian farmers are
already in debt, and because of this few of them are taking their own lives;
hence, bearing the cost of the new technology will not be an easy task for
the farmers.
Technologies need maintenance to keep them running. Thus, such main-
tenance is another additional cost for the farmers. Furthermore, an Indian
farmer holds less than 5 acres of land, so the output will be limited and tak-
ing out all the expenses will be diffcult.
By advance technology in farming, farmers can lose their jobs, and in
India, around 70% of the population depends on agriculture directly or indi-
rectly, so this will push most of the people into poverty.
Poor farmers are denied access. As a result, roughly 70% of farmers have
land that is smaller than 5 acres, making it incredibly diffcult for them to
implement the technology.

7.7 Conclusion
AI complies with an important role in agriculture that moves toward more
automation and acting on real-time applications with an accurate system
in place. Precision agriculture came into picture, instead of traditional
agriculture, with its low cost and advanced tools and equipment. For this,
 Artifcial Intelligence in Agriculture: A Review ◾ 95

applications need to be more robust so that the scope in agriculture lives on

and miracles are made out of it.
Farmers are displaying signs of confdence of a potential future by adopt-
ing cognitive solutions that handle realistic challenges. This leads to people
becoming capable of buying suffcient food. All these applications and
techniques develop technology to make everybody live better on this earth.
The agriculture land is farmer’s legacy, and “farmer is the only person in our
economy who buys everything at retail, sells everything in wholesale, and
pays the freight both ways,” this was the statement by John F Kennedy [9].
If we don’t talk about a farmer, there will be no perfect meaning for agricul-
ture. Farmer is a “Character,” if character adds on with intelligence—that will
be the goal of true and future Agriculture.

References
[1] Marr, D. 1977. Artifcial Intelligence: A Personal View. Artifcial Intelligence,
9(1): 37–48. doi:10.1016/0004-3702(77)90013-3
[2] www.businessinsider.com/artifcial-intelligence?IR=T
[3] https://builtin.com/artifcial-intelligence
[4] Dharmaraj, V., and Vijayanand, C. 2018. Artifcial Intelligence (AI) in
Agriculture. International Journal of Current Microbiology and Applied
Sciences, 7(12): 2122–2128. doi:10.20546/ijcmas.2018.712.241
[5] Sharma, R., Kamble, S. S., Gunasekaran, A., et al. 2020. A Systematic Literature
Review on Machine Learning Applications for Sustainable Agriculture Supply
Chain Performance. Computers and Operations Research, 119: 104926.
[6] Shaf, U., Mumtaz, R., García-Nieto, J., Hassan, S. A., Zaidi, S. A. R., and Iqbal,
N. 2019. Precision Agriculture Techniques and Practices: From Considerations
to Applications. Sensors, 19(17): 3796. doi:10.3390/s19173796
[7] Pandey, A., and Srivastava, M. 2019. A Study of AI Agents in Agriculture—
Present Application & Impact. Amity Journal of Computational Sciences (AJCS),
3(2), ISSN: 2456–6616 (Online).
[8] Nawaz, A. S. N., Nadaf, H. A., Kareem, A. M., and Nagaraja, H. 2020.
Application of Artifcial Intelligence in Agriculture-Pros and Cons. Vigyan
Varta, 1(8): 22–25.
[9] www.fnancialexpress.com/opinion/
if-farmers-remain-poor-so-will-the-country-here-is-why-and-how/1035532/
Chapter 8

Machine Learning and

Artifcial Intelligence-Based
Tools in Digital Marketing:
An Integrated Approach
Preetha Mary George, Sanjeev Ganguly,
and Venkat Reddy Yasa

Contents
8.1 Introduction ..............................................................................................98
8.2 Literature Review .....................................................................................99
8.2.1 Key Concepts ...............................................................................99
8.3 Methodology .......................................................................................... 101
8.4 Findings .................................................................................................. 102
8.4.1 The Perceptions of the Management about the
Analysis Tools ............................................................................. 102
8.4.2 Awareness Among Managers/Top Management with
Various Terms Related to AI and Ml.......................................... 103
8.4.3 Tools Used by Managers (Based on AI and ML) and How to
Adopt ML in Digital Marketing ..................................................104
8.4.4 Role of AI- and ML-Based Analytical Tools in Taking
Decisions and in Setting Up the DM Strategies ........................104
8.5 Conclusion ..............................................................................................105
References .......................................................................................................106

DOI: 10.4324/9781003206316-8 97
98 ◾ Machine Learning for Business Analytics

8.1 Introduction
Artifcial intelligence (AI) and machine learning (ML)-based algorithms have
transformed the business and in different working strategies of the organi-
zations in each and every sector. The technology has improved rapidly in
recent years, which has created a completely dynamic digital marketing envi-
ronment. About 80% of available data on the internet (Big Data) has been
generated in just few years [1]. The interaction and communication among
the companies and its customers has now become very easy due to the use
of such AI-based technology. Also when it comes to the marketing perspec-
tive, AI has completely made it unique and more individualized (digital mar-
keting based on AI is more focused to address the client individually).
Data present on social media sites, search engines, and advertising and
shopping sites is huge, and thus, companies are not investing more and
more to improve the ML ability and to improve the digital marketing strate-
gies. Many organizations from various sectors nowadays use online market-
ing so as to boost business and to build a good marketing strategy [2].
We know that it is very diffcult to interpret such huge data and to draw
insights. It will take more time and could possibly involve many errors,
which can lead to draw wrong conclusions. So many analytical tools have
been used by the companies these days, so as to increase the speed of
work, to remove the chances of error, to optimize and systemize the pro-
cesses, and to reduce human intervention—automate the work.
The technologically advanced ML tools are now being used by most
of the sectors so as to learn and use the past activities and historical data of
the customers and thus help in the digital marketing strategies of the
organizations.
One of the major reasons why companies are trying to shift to AI- and
ML-based analytical models is because of the complexity in using the basic
statistical tools. The basic analytical tools used by the managers were com-
plex; very little was known about these tools and also human intervention
was more as compared to AI-based analytical tools. But when it comes
to the overall usage of this technology in digital marketing, it is very less,
that is, the use of AI and ML is still at its starting stages only. This research
mainly explores that—up to what extent the managers or top management
are aware about these tools. The perceptions of the company’s top man-
agement concern were about AI and ML and these tools; the degree to
which they are currently using these tools; and how to adopt and use
 Digital Marketing: An Integrated Approach ◾ 99

these AI-based tools in their strategic decisions, especially in digital

marketing [3].

8.2 Literature Review

8.2.1 Key Concepts

Artifcial Intelligence—Human intelligence is simulated to machines,

which makes the machine to respond or act intelligently as a human.
In simple words, we can say that AI is basically making a machine to
behave, respond, or interact (in any situation) in the same way as humans
do. An artifcially intelligent machine possesses all the similar traits to
human mind so as to respond in a similar fashion (as humans do). A per-
fect example is the face recognition function in mobile phones.
Machine Learning—ML is particularly a part/subset of AI, where system
learns and automatically with no human intervention involved. ML basi-
cally involves developing computer programs, which can use the data
easily to learn for themselves. Some of the techniques of ML are super-
vised ML, unsupervised ML, semi-supervised ML, and reinforced ML [1].
Big Data—It is a feld which deals with the analyzing, revealing various
information, trends, and interpretation of extremely large amount of data
sets that are practically impossible to be dealt by the traditional methods
of analyzing data. Examples include users comments, reactions, reviews
on social media about any particular company, brand, or a product [4].

Previously, the strategic decisions made by the organizations were depen-

dent on the data gathered internally to the organizations which were the
past data only. But nowadays, the decisions are taken while predicting
the future, so as to gain a competitive advantage in the market. By the
availability of new AI-based analytical tools and huge data (Big Data), it
has become more easy for the companies to analyze and know about the
consumer purchase behavior, reviews about any particular product and thus
to make more effective, long-lasting strategies, and faster decisions. Also,
it is easier for the companies to know about the success rate of the current
strategies implemented by them in the feld of digital marketing. Selection
of the analytical tool is totally dependent on the type and the complexity of
the business.
100 ◾ Machine Learning for Business Analytics

These AI-based analytical tools in digital marketing can:

i. easily access Big Data,

ii. clearly structure the data,
iii. help in making effective and faster decisions,
iv. target individual customer’s needs,
v. clearly indicate the company success factors,
vi. help in engaging with the customers more easily,
vii. help the managers to target new customers easily, and
viii. adapt the digital advancement very easily.

New integrated skills actually boost system performance constantly. ML con-

sists of procedures through which a computer learns (similarly as humans).
One of the best examples of AI is a chatbot on a website. This not only will
boost the ability of client support department but also increases automation
with no human intervention. Various studies have also claimed and showed
that data mining, ML, and text mining are used to analyze the behavior of
customer on a website. Some literatures showed that many new methods
have been developed, to extract the information about the user behavior
and intentions—just by analyzing their posts, comments, and their reactions.
These techniques help the marketers and managers to get a deeper insight
about users and to develop a better understanding about their customers,
which helps them to make their advertisement content accordingly for dif-
ferent segments [3]. Getting insights about the users’ intention from online
platforms or social media sites has very good response for digital market-
ing feld. Conditional random feld—which is an advanced graphical model,
used for sequence data has now been developed, and thus, these things
have taken the ML models to greater heights, in digital marketing.
While doing the literature review, we came to know about various studies
which showed some of the biggest advantages of analytical tools (AI based)
in digital marketing. Also with the introduction of new tools of analysis,
there are both challenges and opportunities in the area of online/digital
marketing.
Finally we can say that:

i. By the use of these tools, machine performs optimally and steadily.

ii. Calculation is much faster while performing analysis, and thus, these
tools enables us to make better and faster decisions.
iii. Complex analysis can be done very easily with vey less chance of error.
 Digital Marketing: An Integrated Approach ◾ 101

iv. Managers also get insights about those users, areas, and segments (for
digital marketing), which are unexplored by the organization.
v. Routine activities can be optimized easily, and there is less or no
chances of error as human intervention is not there while performing
with these AI- and ML-based analytical tools—research aim.

The aim of this research study is to identify the reasons—why these

AI-based tools are not used frequently at a good speed by the organizations.
Secondary aim of the study is:

i. To know about the perceptions of the management about these AI- and
ML-based analytical tools.
ii. To identify the level of awareness among managers, with various terms
related to AI and ML.
iii. To know about the AI- and ML-based tools used by the managers and
how to adopt ML in digital marketing.
iv. To know about the role these tools play in decision-making and in set-
ting up digital marketing strategies.

8.3 Methodology
Ongoing through various research papers, we found that most of them have
used questionnaires and conducted in depth interviews in order to collect the
responses from the respondents. Convenient sampling was used for select-
ing the sample size, and most of the respondents were the higher ranked
employees (departmental heads) or top management employees/managers
in the organizations (organizations which are involved in digital marketing).
Employees at executive level were also selected by the researchers in some
of the studies. For in-depth interviews, experts from marketing and strategy
development department were chosen as they have full knowledge about
which analytical tools company is using for analyzing the online data.
Questions in most of the studies generally include: (i) general questions—
just to collect basic information about the respondent and to know whether
he/she is eligible to answer the questionnaire. (ii) Some questions were
included to check the usage of analytical tools in the organizations, so as to
know about managers’ perceptions about these AI- and ML-based analyti-
cal tools. (iii) Awareness level questions—just to check the level of aware-
ness with various terms associated to AI and ML. (iv) Practical applications
102 ◾ Machine Learning for Business Analytics

questions—they were included just to collect the respondent’s opinion

about—how these tools help in decision-making, practical applications of ML
in digital marketing and in setting up digital marketing strategies. Also in the
questionnaire, the questions included were a mix of rating scale questions,
closed-ended questions, and open-ended questions. Most of the questions
were closed-ended, while a few questions were based on the Likert scale.
But when it comes to open-ended questions, we observed that most of the
researchers have only included one open-ended question or maximum two
open-ended questions. As analyzing these open-ended responses is a diff-
cult task, they just kept these questions to get only the respondent’s opinion
about any particular area or topic.
So, these types of questions were asked by the researchers, and then con-
clusions were drawn based on the responses.

8.4 Findings
Based on the information obtained, the fndings of the research paper are as
follows.

8.4.1 The Perceptions of the Management

about the Analysis Tools
The main intention is to fnd out how respondents distinguish market analy-
sis and analytical tools. The intention was also to know how to use the tools
for their online marketing processes. The questions that were asked to the
respondents were broadly on the basis of the market analysis that they do
for planning process and implementing online strategies for marketing. All
the information (activities) on the internet is measurable with the help of
digital technologies, and this information is collected in making effective
decisions for their growth. Managers are very much aware about it, and they
also use analytical tools in order to measure and analyze online content.
From the responses, it is clear that analysis is important as it aids in provid-
ing sources of information. These information sources are then again used
to prepare and implement the digital marketing strategies. There are many
analysis tools present in the markets that can be used for improving DM
strategies and to gain advantage over the competitors. Some of these online
analytical tools are Google Analytics, Google Data Studio, Facebook insights,
 Digital Marketing: An Integrated Approach ◾ 103

Facebook ad manager, etc. [5]. The aforementioned tools can be integrated

for the preparation of online marketing strategies. Research papers suggest
that the list of analytical tools is used mainly by agencies, media houses,
and advertisers. The usage of analytical tools is completely based on the
business needs. From the recent research papers, we get to know that a
large section of people believe that it is important to use data analytical
tools in digital marketing, and Big Data plays a vital role in online market-
ing. Marketing managers strongly believe that in today’s business, ML tools
play a major role.

8.4.2 Awareness Among Managers/Top Management

with Various Terms Related to AI and Ml
From research papers, we get to know that respondents have rough knowl-
edge about AI and Ml and also Big Data. They understand Big Data as a
huge amount of information than can further be processed. People also feel
that AI and ML along with Big Data create competitive advantage and is the
future of digital marketing [4]. A convenient number of people think that
data management is important and additional education is required because
one needs to have sound knowledge to embrace the opportunities in this
feld. From the research papers, it is clear that respondents use analytical
tools to extract useful information from sales reports and historical market-
ing campaigns. Few gather data from third-party sources and sometimes
free available database. People also feel that AI and Ml and Big Data usage
in digital marketing have few obstacles, and they are:

(i) Most people think that planning and implementing platform data man-
agement is extremely high.
(ii) Implementation of data-oriented approaches is time-consuming.
(iii) Digital marketing is a relatively new platform, and integrating Big Data
requires a lot of information; this puts pressure on educating and train-
ing employees.

When it comes to AI and ML, research papers clearly say that people are
aware about the terms and they feel that these are applicable in present
time. People identifed ML as a kind of process that requires huge amount
of information for any machine to learn and adapt. Many people also agreed
that AI and ML have huge potential in this technology-based DM feld.
104 ◾ Machine Learning for Business Analytics

The process basically involves working with huge/large data (gathered

from websites). Respondents feel that the analysis part will mostly be done
by software and machines, while strategic decision-making and creative
work will be carried out by humans.

8.4.3 Tools Used by Managers (Based on AI and ML)

and How to Adopt ML in Digital Marketing
Research papers gave us an idea that people have basic knowledge about
the AI- and MI-integrated platforms such as Google Data Studio, YouTube,
Google ad words, Facebook ad manager, and Mailchimp. People also know
about the automatic text translators, automatic AI-based cars, and graphics
software/card. We have gone through handful number of research papers,
and we got to know that ML will affect some of the areas in online market-
ing/DM [6, 7]:

i. ML will affect advertising and also the management of advertising

campaigns.
ii. ML will automate the reporting process.
iii. ML will reduce human intervention in the areas where partial automa-
tion is required for example—email or chat communication.

Above are the mentioned areas discussed where ML will minimize human
effort, but areas of creative process and building and maintaining rela-
tionship with partners will limit ML usage. ML-integrated softwares can
develop ideas and also draw from pictures, but the result is not close to
what humans deliver, particularly because in machines, emotional factor is
missing.
People feel that machines are likely to replace the humans in various sec-
tors, but there are still many sectors where it will be diffcult for machines to
control the situations and manage strategically [8].

8.4.4 Role of AI- and ML-Based Analytical Tools in Taking

Decisions and in Setting Up the DM Strategies
Research papers suggest that the framework that aims at encouraging the
use and adoption of these ML- and AI-integrated analytical tools in digital
marketing comprises two things/factors:
 Digital Marketing: An Integrated Approach ◾ 105

i. Management and organizational culture that will create the atmosphere

where the projects will be carried out and successfully fnalized.
ii. A map of the project containing four phases and they are:
a. Management support—without top-level managers’ support, AI and
Ml tools cannot be utilized properly. Top-level employees should frst
lead and must tell the importance of analytical tools and its contribu-
tion toward business success [9].
b. Out-of-the-box solution—a company’s culture should motivate
employees to use analytical tools for innovative solutions.
c. Technical infrastructure—every company should have a team that
has technical knowledge for everyday operational needs. It is an
advantage for a company that has an IT team or has good relation-
ship with technical partners and developers.
d. Focus on data analysis—it is important for the companies to do a
detailed analysis of the data available as it will help them to deploy
their amenities in a better way to gain competitive advantage
through their DM strategies and will also help them to boost internal
management [5, 10].

8.5 Conclusion
After reading research papers in depth, we get to know that market analysis
is important, as it helps to take better strategic decisions. It is also confrmed
that people depend on market analysis tools and analytical tools before
making a digital marketing plan. These analytical tools, used by organi-
zations, have boosted their growth trajectory. Upon reading the research
paper, we got to know that analytical tools can help companies in certain
ways, and they are

a. help in understanding competitor activities and their strategy;

b. aid in important internal data into third-party analytical tools;
c. aid in visualizing of data;
d. aid in quick access to data and information; and
e. help in tracking real-time data for ongoing campaign.

After understanding the research papers, we conclude that the usage of AI

and Ml will be of great in the feld of digital marketing. It will also help
companies to use quality data and process automation. There is no doubt
106 ◾ Machine Learning for Business Analytics

that AI and ML have the place of digital marketing. Despite its hurdles,
marketing managers believe that AI and ML are the future of DM. But in the
current scenario, it is clear that there is still room for adoption of advanced
technology in the workplace despite its availability.

References
[1] Ma Liye, and Baohong Sun, “Machine Learning and AI in Marketing–
Connecting Computing Power to Human Insights”, International Journal of
Research in Marketing, Vol. 37, No. 3, pp. 481–504, 2020.
[2] Geng Cui, Man Leung Wong and Hon-Kwong Lui, “Machine Learning for
Direct Marketing Response Models—Bayesian Networks with Evolutionary
Programming”, Management Science, Vol. 52, No. 4, 1 April 2006.
[3] Hair Jr, Joseph F., and Marko Sarstedt, “Data, Measurement, and Causal
Inferences in Machine Learning: Opportunities and Challenges for Marketing”,
Journal of Marketing Theory and Practice, Vol. 29, No. 1, pp. 65–77, 2021.
[4] Pal Sundsoy, Johannes Bjelland, Asif M. Iqbal, Alex Sandy Pentland and Yves-
Alexandre de Montjoye, “Big Data-Driven Marketing: How Machine Learning
Outperforms Marketers’ Gut Feeling”, Social Computing, Behavioural-Cultural
Modelling and Prediction, MIT Open Access Articles, 2014, pp. 367–374.
[5] Marketing Analytics with AI Complete Guide, https://research.aimultiple.com/
marketing-analytics/
[6] Joni Salminen, Yognathan Vighnesh, Juan Corporan, Bernard J. Jansen and
Soon-Gyo Jung, “Machine Learning Approach to Auto Tagging Online Content
for Content Marketing Effciency—A Comparative Analysis between Methods
and Content Type”, University of Bradford, 2019, Elsevier.
[7] Tim Mackey, Janani Kalyanam, Josh Klugman, Ella Kuzmenko and Rasmi
Gupta, “Solutions to Detect, Classify and Report Illicit Online Marketing and
Sales of Controlled Substances via Twitter—Using Machine Learning and Web
Forensics to Combat Digital Opioid Access”, JMIR Publications, Vol. 20, No. 4,
27 April 2018.
[8] Vinicius Andrade Brei, “Machine Learning in Marketing—Overview, Learning
Strategies, Applications and Future Developments”, Foundations and Trends in
Marketing, Vol. 14, No. 3, pp. 173–236, 31 August 2020.
[9] Louis Columbus, “10 Ways AI and Machine Learning Are Improving Marketing
in 2021”, Forbes, 21 February 2021.
[10] Myron Monets, “How Artifcial Intelligence and Machine Learning Can Be
Used in Marketing”, CURRATI—Editor of Chaos, 23 April 2018, https://curatti.
com/ai-machine-learning-marketing/
Chapter 9

Application of Artifcial
Intelligence in Market
Knowledge and B2B
Marketing Co-creation
H. Raghupathi, Debdutta Choudhury,
and Cynthia Jabbour Sfeir

Contents
9.1 Introduction ............................................................................................108
9.2 Literature Review ...................................................................................108
9.3 Artifcial Intelligence and B2B Market ..................................................109
9.4 Challenges of Artifcial Intelligence ....................................................... 110
9.5 Methodology .......................................................................................... 111
9.6 Findings .................................................................................................. 112
9.6.1 AI in Co-creation of B2B ........................................................... 112
9.6.2 B2B Funnel ................................................................................. 112
9.6.3 Market Knowledge in B2B ......................................................... 112
9.7 Conclusion .............................................................................................. 113
References ....................................................................................................... 114

DOI: 10.4324/9781003206316-9 107

108 ◾ Machine Learning for Business Analytics

9.1 Introduction
Businesses have often benefted from technological advances because they
have provided new ways to reach out to consumers. Artifcial intelligence
(AI) is one of the most important innovations of our time, and it is caus-
ing quite a stir in the modern world [1, 2]. AI in B2B marketing is here to
change the way people communicate with brands, content, and services
because of its marketing ability.
AI in B2B marketing will beneft not only companies, but it will also ben-
eft consumers by inspiring them and providing them with more than they
can imagine [3].
Effect of AI on business:

◾ Centered on cognitive technology outcomes and makes faster business

decisions.
◾ AI programs can minimize errors and “human error” if they are well
set up.
◾ Use data to anticipate consumer needs and have a more personalized
experience.

AI marketing based on collection of data, study, as well as additional audi-

ence or economic dynamics fndings can affect marketing activities. AI
is often used in marketing campaigns where speed is signifcant. AI can
also provide industrial frms with various forms of market intelligence that
are essential for B2B marketing [4]. Learning a person’s surroundings and
behaving in order to accomplish goals are only a couple of the mechanisms
involved in successful intelligence adaptation [1].
In this paper, we briefy overview that the theoretical changes of mar-
keting implemented have resulted from technological advancement in the
digital environment, especially in the B2B context [5].

9.2 Literature Review

Technology is evolving at a rapid pace, and organizations are keen on
leveraging this technological advancement in order to help generate more
revenue and propel the business forward. B2B organizations however have
relied heavily on the conventional mode of communication, the communica-
tion takes place through multiple organizing interactions between the buyer
 AI in Market Knowledge and B2B Marketing ◾ 109

and seller. These interactions are meant to build trust factor for the longevity.
The agenda is to infuence the buying behavior of customer with the help of
comprehensive communication strategies [5].
Organizations are entering digital media space to help improve the com-
munication with its prospective clients and social media analytics tools,
and AI is ubiquitous in this predicament [6] AI helps in negating crisis and
improves the ability to co-create. The machine learning (ML) algorithms play
a signifcant role as they help in understanding the sentiment of the supplier
or buyer while conversing through a B2B market place such as IndiaMART.
The ML algorithm uses sentiment analysis to understand the behavior pat-
terns of the parties involved in the process [7].
The e-commerce platforms have reshaped the B2B organization space
and redefned the way they communicate with the buyer and supplier. The
interactions are primarily confned to the chat on the platforms and then
they are taken into telecommunication on as and when required basis. This
provided a huge scope for AI to reduce the work load on people by analyz-
ing the chat process to detect patterns and create FAQ’s for the company [8].
With the world being as it is under this new normal, the B2B organiza-
tion interactions in the physical space has been reduced tremendously which
opens doors for digital B2B interactions and this predominantly happens
over platforms like zoom and Skype, and organizations are creating AI tools
so that buyers can interact with it and customize the service offerings to
their needs [9].

9.3 Artifcial Intelligence and B2B Market

Many areas of B2B marketing are being revolutionized by AI. Gartner (2018)
predicts that by 2020, to drive marketing automation, one-third of all B2B market-
ers would use AI-enabled tools. Thus, using data relating to past user informa-
tion and buying patterns within the B2B business’s website, AI helps to improve
customer connections. This is basically a next-generation customer experience
model. It also opens plenty of expansion and cross-selling possibilities.
Chatbots are another way AI is used in marketing. These bots can assist
with problem-solving, product or service suggestions, and sales support.
Marketers beneft from AI because it analyzes data on customer behavior
faster and more accurately than humans.
Most of us have seen the advantages of AI in our very own experience of
shopping, as in when we use Uber to get a ride, or when we use Netfix to
110 ◾ Machine Learning for Business Analytics

fnd a movie, or when we use Airbnb to book a room. Many conventional

balance sheet and operating income line items are much less predictive of
fnancial viability than effective use of Big Data and data science.
Over the past ten years, marketing departments have used social net-
working, customized website creation, content marketing, account-based
marketing, marketing analytics, targeted advertisement, marketing and sales
coordination, programmatic advertisements, attribution models, and funnel
metrics [10].
Marketers must collect, analyze, and identify data before targeting pos-
sible leads using conventional data generation methods, which take a signif-
cant amount of time. While marketers can generate leads using data from a
variety of websites, CRM systems, and marketing campaigns, there are some
gaps in which marketers fail to understand the buying patterns and interests
of their leads.
AI can easily capture a broader range of data from different sources as
blogs, social media platforms, and contact databases due to its wide cover-
age. AI can close the gap between salespeople and potential customers by
determining the best B2B goals for inbound and outbound marketing cam-
paigns [11, 12].

9.4 Challenges of Artifcial Intelligence

B2B consumers, for example, are often annoyed by irrelevant and intrusive
ads distributed by automated email and social media marketing with AI.
AI-created unsuccessful marketing campaigns have the potential to damage
a company’s brand and credibility.
A solid IT infrastructure is needed for an effective AI-driven marketing
strategy. AI technology processes massive amounts of data, which can be
costly to set up and maintain. They’ll almost certainly need regular updates
and repairs to keep running smoothly. This can be a major stumbling block,
particularly for smaller businesses with limited IT budgets.
Although large corporations can choose to create and operate their
own AI marketing tools, smaller businesses may beneft from cloud-based
solutions.
Even though AI solutions usually have a high return on investment, a
business argument must be made to justify investing in these new technolo-
gies. This is especially challenging in smaller businesses with already tight
budgets. Complex software and high-performance hardware are needed for
AI technology, which is costly to implement and maintain.
 AI in Market Knowledge and B2B Marketing ◾ 111

Organizations no longer have to rely on designing in-house AI solutions,

due to an increasing number of affordable AI vendors. AI marketing technol-
ogies can be applied not only more affordably, but much far more quickly
than in the past.
There is still an AI skills shortage, which can have a signifcant effect
on companies looking to implement AI marketing strategies in-house. To
fll these new vacancies, the existing pool of AI talent is not rising quickly
enough. Although some companies may be able to close the skills gap by
educating current staff, others may need to set aside funds to recruit AI spe-
cialists with a competitive compensation package.
For regulatory purposes, certain companies may be prohibited from stor-
ing data offsite, which means they may be unable to use cloud-based AI
marketing vendors. All companies must take responsibility for ensuring that
AI software is used safely and in a way that benefts their consumers rather
than just their bottom line.

9.5 Methodology
Mixed methods research was followed to analyze the data, where qualitative
research was conducted on North American-based startup company which
provides competitive intelligence services to the customers using AI tech-
nology backing with ML and natural language understanding (NLU). Senior
employees from various organizational levels are participated on semi-
structured interviews with a time slot of half an hour [1]. Further quantitative
research used panel data and common system method of moments caused
by an independent variable that is related to the error word [5].
Population: Managerial level employees from various departments includ-
ing sales, customer service, marketing, and IT-related people participated on
the in-depth semi-structured interviews [1].
Sample Size: A startup company employees based on North America are
considered as the sample size where they service competitive intelligence
services and corporate sales to the clients using AI.
Data Sources: The initial step was data drafting on B2B sales and modi-
fying with the suggestions from the colleagues before conducting the
interviews. After conducting couple of interviews and taking feedback on
questionnaire further modifed based on the suggestions, this process was
repeated two times, and after the fnal best questionnaire, in-depth inter-
views are conducted to analyze the data on B2B sales and value creation by
AI in B2B marketing [13]. The complete interviews time was 7.25 hours with
112 ◾ Machine Learning for Business Analytics

an average time of 30 minutes for each, whereas the range is between 16

and 40 minutes on overall [1].
Interviewer wrote a summarized memo and submitted the audio record-
ings transcribed using professional service of Microsoft word. These summa-
rized memos and audio transcripts are the data set of this study [1].

9.6 Findings
9.6.1 AI in Co-creation of B2B
Initially, when start-up frms want to interact with its potential clients, they
develop an AI application (basically a bot) [6, 7]. This AI bot follows the below
process to incorporate sales of IT products and services is shown in Figure 9.1.

9.6.2 B2B Funnel

There are seven steps in AI as mentioned below [8, 9, 14]. This method has
been made easier thanks to AI. The following sections detail how AI infu-
enced B2B revenue is shown in Table 9.1.

9.6.3 Market Knowledge in B2B

Only when various types of AI frameworks and inputs from various
types of industry expertise are combined can high-quality innovation be
produced [2, 15]. Customer experience, consumer knowledge, and exter-
nal business knowledge are examples of market knowledge [4, 16].

Figure 9.1 AI workfow in co-creation B2B process.

 AI in Market Knowledge and B2B Marketing ◾ 113

Table 9.1 How AI Contributes to Processes in B2B Sales Funnel

Process AI Contribution

Prospecting Lead qualifcation and generation Models

Pre-approach Advertising, retargeting
Approach Contact through digital Agents, content curation
Presentation Prototyping, sentiment analysis
Overcoming objections Competitive intelligence
Closing Dynamic pricing according to client
Follow-up Chatbots, automated workfows

Table 9.2 How AI Contributes to Different Types of Market Knowledge

Type of Marketing Contribution of AI

Knowledge
Customer knowledge • Creating profile of potential customers.
• Structured and unstructured data about customer
attributes like buying behavior and buying process.
• Predictive models for prospect scoring
• Chatbots
User knowledge • Internet of things
• Big Data
• Text analysis
• Social media analysis
External market • Natural language processing
knowledge • Competitive intelligence
• Analyzing press releases and blogs for insights

Table 9.2 gives a brief overview about how AI contributes to market

knowledge.

9.7 Conclusion
Nowadays, due to technological advancement, many segments have started
using AI and ML as their base. AI is showing a huge impact on present
114 ◾ Machine Learning for Business Analytics

marketing activities [1]. The previous results have discussed about how AI is
contributing for co-creation and marketing knowledge in B2B marketing and
sales. AI also reduced the work of marketing or sales professionals in this
vast ocean of B2B marketing [3]. The previous results also show the usage
and importance of AI-based sales funnel. In this summary, the differences
between traditional and AI-based systems are discussed.
By adopting strict concords, dependable templates, and a personalized
approach, value co-creation is encouraged by AI-based providers [4]. The
use of six AI building blocks in B2B marketing was also addressed previ-
ously [17]. Overall, the summary discussed the uses and usefulness of AI in
the B2B feld for co-creation and business awareness.

References
[1] Paschen, J., Kietzmann, J., & Kietzmann, T. C. (2019). Artifcial intelligence
(AI) and its implications for market knowledge in B2B marketing. Journal of
Business & Industrial Marketing, 34(7), 1410–1419.
[2] Paschen, U., Pitt, C., & Kietzmann, J. (2020). Artifcial intelligence: Building
blocks and an innovation typology. Business Horizons, 63(2), 147–155.
[3] Paschen, J., Wilson, M., & Ferreira, J. J. (2020). Collaborative intelligence: How
human and artifcial intelligence create value along the B2B sales funnel.
Business Horizons, 63(3), 403–414.
[4] Leone, D., Schiavone, F., Appio, F. P., & Chiao, B. (2021). How does artif-
cial intelligence enable and enhance value co-creation in industrial markets?
An exploratory case study in the healthcare ecosystem. Journal of Business
Research, 129, 849–859.
[5] Lin, W. L., Yip, N., Ho, J. A., & Sambasivan, M. (2020). The adoption of techno-
logical innovations in a B2B context and its impact on frm performance: An
ethical leadership perspective. Industrial Marketing Management, 89, 61–71.
[6] Murgai, A. (2018). Transforming digital marketing with artifcial intelligence.
International Journal of Latest Technology in Engineering, Management &
Applied Science, 7(4), 259–262.
[7] Farrokhi, A., Shirazi, F., Hajli, N., & Tajvidi, M. (2020). Using artifcial intel-
ligence to detect crisis related to events: Decision making in B2B by artifcial
intelligence. Industrial Marketing Management, 91, 257–273.
[8] Lau, R. Y. (2007). Towards a web services and intelligent agents-based nego-
tiation system for B2B eCommerce. Electronic Commerce Research and
Applications, 6(3), 260–273.
[9] Prior, D. D., & Keränen, J. (2020). Revisiting contemporary issues in B2B mar-
keting: It’s not just about artifcial intelligence. Australasian Marketing Journal
(AMJ), 28(2), 83–89.
 AI in Market Knowledge and B2B Marketing ◾ 115

[10] Huang, M. H., & Rust, R. T. (2018). Artifcial intelligence in service. Journal of
Service Research, 21(2), 155–172.
[11] Wilson, R. D., & Bettis-Outland, H. (2019). Can artifcial neural network mod-
els be used to improve the analysis of B2B marketing research data? Journal of
Business & Industrial Marketing, 35, 495–507.
[12] Blake, M. B. (2002, July). B2B electronic commerce: Where do agents ft in?
In Proceedings of the AAAI-2002 Workshop on Agent Technologies for B2B
E-Commerce, Edmonton, Alberta, Canada.
[13] Martínez-López, F. J., & Casillas, J. (2013). Artifcial intelligence-based systems
applied in industrial marketing: An historical overview, current and future
insights. Industrial Marketing Management, 42(4), 489–495.
[14] Tahvola, K. (2020). Leveraging artifcial intelligence in B2B Markets.
[15] Paschen, J., Kietzmann, J., & Kietzmann, T. C. (2020). Unpacking artifcial
intelligence—How the building blocks of artifcial intelligence (AI) contribute
to creating market knowledge from big data.
[16] Chen, A. P. S., Chansilp, K., Kerdprasop, K., Chuaybamroong, P., Kerdprasop,
N., Shun-Fa, H., . . . Kaium, M. A. (2020). B2B marketing crafts intelligence
commerce: How a Chatbot is designed for the Taiwan Agriculture Service.
International Journal of e-Education, e-Business, e-Management and
e-Learning, 10(2), 114–124.
[17] Vladimirovich, K. M. (2020). Future marketing in B2B segment: Integrating
artifcial intelligence into sales management. International Journal of
Innovative Technologies in Economy, 4(31).
Chapter 10

A Systematic Literature
Review of the Impact of
Artifcial Intelligence on
Customer Experience
M. A. Sikandar, Praveen Kumar Munari, and Meghraj Arli

Contents
10.1 Introduction ............................................................................................ 118
10.2 Conceptual Backgrounds: Literature Review ........................................ 118
10.2.1 Defning Experience .................................................................. 118
10.2.2 Customer Experience Management Defnition ......................... 119
10.2.3 Artifcial Intelligence .................................................................. 119
10.2.4 Customer Personalization...........................................................120
10.3 Customer Experience and Journey ........................................................120
10.3.1 Types of Consumer Experiences ............................................... 121
10.3.2 Artifcial Intelligence and Customer Experiences .....................123
10.4 Research Methodology ...........................................................................123
10.5 Results and Conclusion ..........................................................................124
References .......................................................................................................125

DOI: 10.4324/9781003206316-10 117

118 ◾ Machine Learning for Business Analytics

10.1 Introduction
Artifcial intelligence (AI) is the most signifcant technological advance; these
machine-learnt technologies beneft both businesses and customers. The
journey that a customer takes from pre-purchase to post-purchase can be
defned as the user’s experience (Braun and Carriga, 2017).
Personalization and AI interaction have the potential to bring customers
closer to online retailers (Parkes, 2018). AI is meant to improve customer
experience and thus help businesses adapt to “servitization.”
However, machine learning (ML) does not evolve changes in the path
when online retailing is done (Amazon Annual Report, 2018). It also infu-
ences how customers in the online retailing shop in traditional brick-and-
mortar stores. As a result, the gap between offine and online retailing
(brick-and-mortar stores) is closing (e-commerce) (Silver, 2016).
Department stores, as well as internet sites in the omnipresent
world, could not be regarded as independent bodies. Instead, the two
are advantageous to one another. As a result of this omnichannel view
of commerce, customers’ purchase paths are becoming much more
complicated.
According to Hogg (2018) in an article written for Google, a travel guide
with exits, research, and conversation across the path is identical to a cus-
tomer lifecycle, whenever it is necessary to persuade, customers to choose
and stay with your company instead of turn to a rival (LG, 2019). As a result,
it is diffcult for marketers to create a value offer that is tailored to the cus-
tomer’s unpredictable desires. As a result, businesses must plot their cus-
tomers’ paths to purchase by using their leverage technology advancements
to optimize consumer loyalty in the purchasing decision (Kaci, Patel, and
Prince, 2014). Machine learning (ML) can increase user travel; however, com-
panies must know how these innovations affect the customer perspective in
this ever-changing environment

10.2 Conceptual Backgrounds: Literature Review

10.2.1 Defning Experience
Customer experience refers to a user’s intrinsic and impartial reaction to
interacting directly with a company. Even during purchase, use, and sup-
port of a product, the customer usually initiates direct communication and
 The Impact of AI on Customer Experience ◾ 119

experience. A series of encounters between an individual and a product, a

business, or a feature of its organization that conveys a refex is referred to
as consumer experience. This is a highly personal observation that neces-
sitates the customer’s involvement on many levels. Its assessment is based on
a comparison of the needs of a customer and the sensations obtained from
experiences with the business and its offerings in relation to various multiple
touchpoints or instances of contact (Bolton, Gustafsson, McColl-Kennedy,
Sirianni, and Tse, 2014).
The aforementioned concept of customer engagement is multidimen-
sional, encompassing sensory, cognitive, adaptive, physical, and emotional
elements (DePillis and Sherman, 2019). Secondly, it understands the impor-
tance of both logical and emotional dimensions of customer service.

10.2.2 Customer Experience Management Defnition

In a Seminal Article published in 1998, Joseph Pine II and James H. Gilmore
coined the word “Customer Experience Management.” The writers stressed
the importance of making a lasting impression on the consumer (Bhandari,
Rama, Seth, Niranjan, Chitalia, and Berg, 2017).
Customer experience is a clear refection and fawless implementation
of the emotional link and interactions you intend your consumers should
have with any brand through delivery channels and contact points. Later,
customer interaction management is understood as the method of effciently
managing a buyer’s entire experience with a brand or service, according to
each (Lutz, 2017). The signifcance of incorporating different kinds of cus-
tomer satisfaction through multiple touchpoints was emphasized.

10.2.3 Artifcial Intelligence

Crittenden, Biel, and Lovely (2019) defne AI as the concept of machines
capable of performing activities intelligently; delivering useful, effcient solu-
tions to issues. Individuals also assert that AI is related to other technical
advancements including machines and data science. Computer learners have
the opportunity of computers to learn on their own without the use of com-
prehensive programming, and robots are capable of performing tasks that
humans would normally perform (Amazon Echo, 2019; Amazon Echo Auto,
2019; Amazon Locker, 2019).
Geisel (2018) defnes AI as follows: a truly artifcially intelligent system by
itself will read. We speak of ML, which could shape links while depending
120 ◾ Machine Learning for Business Analytics

on pre-programmed conduct equations and extract meanings. True AI can

increase its strengths or experience through earlier installments, becoming
wiser and much more aware as it progresses (Levy, 2018).
AI may be considered to be “intelligent” and “automated” systems. These
meanings do not seem quite precise and are generally very technical. In the
course of this thesis, to provide a better picture of the implications of AI, I
will explore different IC strategies and developments in sales and economics
(Chamatkar and Butey, 2015).

10.2.4 Customer Personalization

It is important to recognize the actions of customers, as per Chatterjee
(2018) because that knowledge would be used to defne and address a tar-
geted audience adequately in line with your wants and requirements. The
importance of being is emphasized by Google (2015) applying the prin-
ciples of “being rapid.” One of the measures organizations could undertake
when doing operations is to anticipate needs. Firms should look at previ-
ous actions of a customer, can classify them, and include a relevant mes-
sage tailored to the individual. It is important to consider consumer conduct
and provide the consumer with an unfriendly experience. Chatterjee (2018)
discusses a “Consumer Behaviors DNA” algorithm. This algorithm (which
employs pattern mining) assists businesses in identifying various types of
customer behavior.
Companies can use this data to better segment customers and gain a bet-
ter understanding of their needs/preferences. With this understanding and
information about purchasing behavior, businesses can use recommender
systems to better target their customers and personalize their services.

10.3 Customer Experience and Journey

Customer experience is just how consumers view the organization experi-
ences. A good approach from either the customer’s perspective must be pro-
ductive, helpful, and fun. The particular aspect of the customer perception,
a client’s character, restricts its management utility for forecasting and con-
trolling reasons at a specifc moment and place in a particular case. Many
enterprises consider customer satisfaction (CEM) as a customer relationship
management successor (CRM). But interventional integration problems are
becoming even more challenging.
 The Impact of AI on Customer Experience ◾ 121

Figure 10.1 Factors that affect customer satisfaction.

For this analysis, fve main criteria for measuring customer service were
considered in terms of service quality and service retention, according to UCTI,
the University College of Technology and Innovation of Asia Pacifc. With
the aid of Figure 10.1, we can quickly obtain a general summary of customer
satisfaction.
A customer journey map provides you with more information about
the consumer, allowing you to go beyond what you already know. Many
companies regard the customer journey as something observable – the
point at which the client engages with the company. However, this is not
the case, and only accounts for a small portion of the whole consumer
experience. Making a customer journey map forces you to consider the
components of the trip that you don’t see but are as important to the over-
all experience.

10.3.1 Types of Consumer Experiences

By the study of a few research papers like Braun and Carriga’s (2017) philo-
sophical analyses, Schmitt (2011) stated that fve kinds of relations like hear,
feel, perspective, act then relate.
Sense: It’s a sensory value that offers an amazing event stimulus by the
clear response to the fve human senses: consumer sense, hearing sense,
122 ◾ Machine Learning for Business Analytics

consumer feel, tasting sensation, and odor sensing. On the road, for exam-
ple, a Jaguar gives a case of a car responsive experience, this attraction is in
sensual meaning. Design sense. A Ferrari, on the other hand, has a respon-
sive and exciting perceptual value.

i. Feel (affective customer experience): The affective experience calls

on the personal feelings and ideas of consumers to create affective
interactions from a slightly optimistic point of view market behaviors
(e.g. for non-competitive moods, brand, operation, or industrial non-
sustainable food product) to high pleasure and pride emotions (e.g. sus-
tainable market, technology). Feels as though it is, attributes of feeling
that are related to consumer sentiments and emotions. There are rules
of emotional experience that we have to drink a cup of coffee store, the
excitement we’re living cherishing a trip to Paul Krugger’s National Park,
and there are loose feelings that we have to drink a cup of coffee store.
ii. Consider the following (cognitive/creative business performance):
I believe in the mind intending to develop cognitive interactions that
produce outcomes while also involving our clients’ innovative think-
ing. Think; attempts by suspense, excitement, and agitation to involve
consumers’ inductive and deductive thought. The latest technological
merchandise is thought of as campaigns. Think of artistic knowledge
principles that attract people reasoning abilities via the life experience
that resolve issues and cognitive problems (Chamatkar and Butey, 2015).
iii. Physical customer service (Act): This Act is designed to infuence
physical perceptions and lifestyles. Marketing enhances the lives of
consumers by improving physical environments, alternative practices
and relations, as well as new ways of doing things. Ethical methods
to personality development associated with the act are usually motiva-
tors, inspiring, and creative by positive examples, such as celebrities or
well-known players (Chamatkar and Butey, 2015). Act is about standards
of conduct that relate to resistance training, attitudes, and interpersonal
interactions.
iv. Relate (social-identity customer experience): RELATE includes ele-
ments of the selling of senses, feelings, thought, and actions. That being
said, relationships extend far beyond intimate, secret thoughts of the
partner and therefore connect to “personal experiences” and connect
the person to his/her values, other persons, or community (Chamatkar
and Butey, 2015). The connection is about relative values of experi-
ence which call for personal liberalization. Nike’s affection means that
 The Impact of AI on Customer Experience ◾ 123

a customer puts on the rear of the wrist a tattoo to display him/her the
company.

10.3.2 Artifcial Intelligence and Customer Experiences

The ability of AI by simplifying consumer experiences and purchasing
trips is to enhance the importance in the business world. Marketing com-
panies and intelligence experts have to cooperate but not fght to achieve
this (Chamatkar and Butey, 2015). Salespeople should rethink their service
delivery and concentrate on AI to improve confdence or pace in operations,
improve the volume and make better decisions, and create more effcient
customization. When people and sophisticated technology partner, compa-
nies can see their greatest cost savings. Companies need workers to develop
technology and clarify their actions to ensure the effcient use of processors
(Freitas, Gastaud Maçada, Brinkhues, and Zimmermann, 2016).

10.4 Research Methodology

There hasn’t been a lot of research done on the subject of AI in management.
Although AI is also a relatively young feld, the majority of research in this
feld is either continuing or unfnished. However, there is AI research available
in other felds that contributes to our understanding of AI in the industry.
The research I conducted for this thesis was in the feld of marketing, spe-
cifcally the consumer experience and journey. To evaluate the infuence of AI
on the customer journey, we conducted a literature review. This analysis takes
into account previous research on the topic. The objective of this research
work is to allow the researcher to understand the theory by research ques-
tions. A lot of information available inquiries in an internet source are part of
the research quest. Keywords used in publications contain (a) customer expe-
rience; (b) e-commerce (Piyush, Choudhury, and Kumar, 2016) and (c) AI.
The following were the conditions for inclusion: English language; busi-
ness, management, and accounting and to reduce the number of results, a
separate search was conducted for more general knowledge on AI and the
user path, using different inclusion criteria such as region and time. In addi-
tion, I used Amazon.com as a case study and analyzed it using blogs, press
releases, and news stories. Companies should use Amazon as a best practice
example for improving their customers’ experiences with technology like AI,
according to this case study (Viktor, 2019).
124 ◾ Machine Learning for Business Analytics

10.5 Results and Conclusion

Frameworks to demonstrate the effect of classifcation models and chatbots
can have on service quality, as well as to map such deep learning tech-
niques through the customer journey, as previously discussed. The activities
of customer experiences are as shown in Figure 10.2.

Figure 10.2 Customer activities.

 The Impact of AI on Customer Experience ◾ 125

However, understanding the consumers’ behavior and desires are critical

to do so. Consumer experience is affected by both practical and psychologi-
cal factors, and it is important to gain a better understanding of customer
behavior by determining how AI techniques affect these. Customer person-
alization is possible with recommender systems, while customer experience
can be improved with chatbots. The following is how the customer experi-
ence is defned: The customer experience is made up of the interpretations
that buyers have while communicating with a business.
Customers’ qualitative reactions would be favorably affected if AI could
be used to improve interactions at different touchpoints. Many of the better
experiences from the enhanced shopping journey are brought together, from
the beginning to the end of the transaction.
The impact of AI on the consumer experience during the shopping expe-
rience was investigated in this report. This has been discussed as part of a
broader context focused on previous research. With the increasing signifcance
of providing customers with a special experience, literature that illustrates the
impact of technological advancements on this experience is required. Piyush,
Choudhury, and Kumar (2016) suggest a structure for new value development
channels in the digital era. Automation, individualization, ambient embedded-
ness, interaction, openness, and control are some of these outlets.

References
Amazon Annual Report. (2018). Amazon Annual Report 2018. Retrieved on 02–06–
2019 from https://ir.aboutamazon.com/static-fles/0f9e36b1-7e1e-4b52-be17-
145dc9d8b5ec
Amazon Echo. (2019). Amazon Echo. Retrieved on 02–06–2019 from www.amazon.
com/all-new-amazon-echo-speaker-with-wif-alexa-dark-charcoal/dp/B06XCM9LJ4
Amazon Echo Auto. (2019). Amazon Echo Auto. Retrieved on 02–06–2019 from
www.amazon.com/Introducing-Echo-Auto-frst-your/dp/B0753K4CWG
Amazon Locker. (2019). Amazon Locker. Retrieved on 01–06–2019 from www.
amazon.com/b/ref=amb_link_366591722_2?_encoding=UTF8&node=64426000
11&tag=bisafetynet2-20
Bhandari, A., Rama, K., Seth, N., Niranjan, N., Chitalia, P., Berg, S. (2017). Toward
an Effcient Method of Modelling “Next Best Action” for Digital Buyer’s
Journey in B2B. In: International Conference on Machine Learning and
Data Mining in Pattern Recognition, MLDM 2017, Lecture Notes in Computer
Science, 10358, Springer, Cham, pp. 107–116.
Bolton, R.N., Gustafsson, A., McColl-Kennedy, J., Sirianni, N.J., Tse, D.K. (2014).
Small Details That Make Big Differences: A Radical Approach to Consumption
Experience as a Firm’s Differentiating Strategy. Journal of Service
Management, 25(2), pp. 253–274.
126 ◾ Machine Learning for Business Analytics

Braun, A., Carriga, G. (2017). Consumer Journey Analytics in the Context of Data
Privacy and Ethics. Digital Marketplaces Unleashed, pp. 663–674, Springer
Nature Switzerland.
Chamatkar, A.J., Butey, P.K. (2015). Implementation of Different Data Mining
Algorithms with Neural Network. In: 2015 International Conference on
Computing Communication Control and Automation, 26–27 Feb., Pune,
India, IEEE, pp. 374–378.
Chatterjee, R. (2018). Amazon’s 2 Most Powerful Rivals Just Decided to Team Up.
Business Insider. Retrieved on 02–06–2019 from www.businessinsider.com/r-
walmart-microsoft-in-partnership-to-use-cloud-tech-2018-7?international=true&
r=US&IR=T
Crittenden, W.F., Biel, I.K., Lovely, W.A. (2018). Embracing Digitalization: Student
Learning and New Technologies. Journal of Marketing Education, 21(1),
pp. 5–14.
DePillis, L., Sherman, I. (2019). Amazon’s Extraordinary Evolution: A Timeline.
CNN. Retrieved on 23–05–2019 from https://edition.cnn.com/interac-
tive/2018/10/business/amazon-history-timeline/index.html
Freitas, J.C., Gastaud Maçada, A.C., Brinkhues, R., Zimmermann, G. (2016). Digital
Capabilities as Driver to Digital Business Performance. In: 22nd Americas
Conference on Information Systems, San Diego, USA, pp. 1–5.
Geisel, A. (2018). The Current and Future Impact of Artifcial Intelligence on
Business. International Journal of Scientifc & Technology Research, 7(5),
pp. 116–122.
Google. (2015). Micro-Moments: Your Guide to Winning the Shift to Mobile. Think
with Google. Retrieved on 28–06–2019 from www.thinkwithgoogle.com/
marketing-resources/micro-moments/micromoments-guide-pdf-download/
Hogg, S. (2018). Customer Journey Mapping: The Path to Loyalty. Think
with Google. Retrieved on 28–06–2019 from www.thinkwithgoogle.com/
marketing-resources/experience-design/customer-journey-mapping/
Kaci, S., Patel, N., Prince, V. (2014). From NL Preference Expressions to
Comparative Preference Statements: A Preliminary Study in Eliciting
Preferences for Customised Decision Support.
Levy, S. (2018). Inside Amazon’s Artifcial Intelligence Flywheel. Wired. Retrieved
on 02–06–2019 from www.wired.com/story/amazon-artifcial-intelligence-
fywheel/
LG. (2019). LG Fridge Alexa Amazon. Retrieved on 02–06–2019 from www.lg.com/
us/refrigerators/lg-LNXS30996D-door-in-door
Lutz, A. (2017). Amazon Is Offcially Buying Whole Foods—Here’s Everything That
Will Change for Customers. Business Insider. Retrieved on 23–05–2019 from
www.businessinsider.com/amazon-buys-whole-foods-changes-2017-8?internati
onal=true&r=US&IR=T
Parkes, T. (2018). How One Retailer Is Shifting to an AI-First Mentality. Think
with Google. Retrieved on 28–06–2019 from www.thinkwithgoogle.com/
marketing-resources/experience-design/1-800-fowers-voice-assistants/
 The Impact of AI on Customer Experience ◾ 127

Piyush, N., Choudhury, T., Kumar, P. (2016). Conversational Commerce a New

Era of E-Business. In: 2016 International Conference System Modelling &
Advancement in Research.
Schmitt, B. (2011). Experience Marketing: Concepts, Frameworks and Consumer
Insights. Foundations and Trends in Marketing, 5(2), pp. 55–112.
Silver, C. (2016). Amazon Announces No-Line Retail Shopping Experience with
Amazon Go. Forbes. Retrieved on 23–05–2019 from www.forbes.com/sites/
curtissilver/2016/12/05/amazon-announces-no-line-retail-shopping-experience-
with-amazon-go/#f7cf2d12326e
Viktor, R. (2019). Using Adversarial Training to Recognize Speakers’ Emotions.
Amazon Alexa Blogs. Retrieved on 22–05–2019 from https://developer.
amazon.com/blogs/alexa/post/2d8c2128-eec9–44cc-9274-444940eb0a4d/
using-adversarial-training-to-recognize-speakers-emotions
Chapter 11

The Impact of Artifcial

Intelligence on Customer
Experience and the
Purchasing Process
Laxmi Shaw, Megha Mankal, and Chinnapani Kiran Kumar

Contents
11.1 Introduction ..........................................................................................130
11.2 The Customer Experience: A Literature Review .................................130
11.2.1 Concepts and Defnitions......................................................... 131
11.2.2 The Dimensions Included in Customer Experience ............... 131
11.2.3 The Customer Experience Measurement.................................132
11.3 Artifcial Intelligence in the Public Sector ...........................................132
11.4 Artifcial Intelligence in Retail Sector ..................................................133
11.5 Artifcial Intelligence in B2B E-commerce ..........................................133
11.6 Challenges of Artifcial Intelligence .....................................................134
11.7 Benefts of a Satisfed Customer ..........................................................134
11.8 Research Methodology ......................................................................... 135
11.8.1 Population................................................................................. 135
11.8.2 Sample Size............................................................................... 135
11.8.3 Data Sources ............................................................................. 135
11.8.4 Conceptual Model ....................................................................136
11.9 Discussion of the Findings...................................................................137
11.10 Conclusion and Future Work ............................................................... 141
References ....................................................................................................... 142

DOI: 10.4324/9781003206316-11 129

130 ◾ Machine Learning for Business Analytics

11.1 Introduction
The concept mainly focusses on implementing artifcial intelligence in
various felds especially in business to customer segment to enhance the
customer experience. Artifcial intelligence (AI) plays a signifcant role for
customer’s engagement and customer’s buying journey.
The role of AI throughout customer’s buying journey is at different phases
like stage of awareness, phase of consideration, process of purchasing, and
support phase. Customer experience can be explained with two important
dimensions namely personalized customer service and after-sales customer
support. Due to globalization and industrial revolution, fulflling customer
expectations which are at higher level has become challenging task for all
industrial segments [1].
AI has gone through various innovations by introducing various platforms
like augmented reality, virtual reality and mixed reality. The mentioned plat-
forms are having wide applications especially in customer’s buying journey.
It will respond and provide services to customer based on demographics
and lifestyle [2].
Chatbots are almost utilized as artifcial salesperson during the pur-
chasing process, by providing appropriate information of products to the
customer. Customer purchase journey geos through three major impor-
tant stages of transactions namely pre-transaction, transaction, and post-
transaction [2].
The aim of our review is to identify the factors that have greater infuence
on customer experience, customer’s buying journey using various innova-
tions, and applications of AI.
In the coming discussions, we elaborate theoretical aspects of customer
experience and customer buying journey. Also, we would like to discuss on
defnitions and concepts of applications and innovations of AI and thereafter
to discuss a separate analysis for both customer experience and consumer
buying journey using applications of AI.

11.2 The Customer Experience: A Literature Review

The signifcance of customer experience and customer buying journey is
well cited in the literature. Also, how AI infuences the customer buying
journey and resulting to achieve greater customer loyalty and proft is well
explained [1].
 Impact of AI on Buyer Experience ◾ 131

The “digital revolution” has fundamentally changed the consumer

experience over the last 20 years. In his book Becoming Digital, Nicholas
Negroponte (1995) [3] described the transition from “atoms” to “bits” to
explain digital technologies and their benefts [2].

11.2.1 Concepts and Defnitions

Customer experience can be explained in various ways of approach; among
them, other prominent customers typically like friends and family are also
causes for differences in one’s own customer experience. The research also
identifed there are different customer factors discussed in literature. The
prominent factors include social factors, task factors, time factors, product
involvement, and fnancial resources [4].
The literature proposes defnitions for customer experience dimensions
such as customer service and after-sales customer care [1].
The concepts of augmented reality, virtual reality, mixed reality are
explained well in the discussion. And the applications like virtual assistants,
chatbots, and robots are well identifed and explained in the discussion [2].
Drawing from those concepts and defnitions, we suggest defning cus-
tomer experience as “[t]he experience which physically and emotionally
connected with customer throughout the customer buying journey across
various phases of purchase process.”

11.2.2 The Dimensions Included in Customer Experience

In addition to the two dimensions of customer experience mentioned earlier
namely customer service and after-sales customer support, few other multi-
dimensions for customer experience are identifed.
The major dimensions of customer experience identifed are SERVQUAL
model, eTailQ Experience, E-S-QUAL, consumption experience model,
customer experience model, conceptual model of customer experience, and
fnally, EXQ framework [5].
All the above dimension models of customer experience have sub-dimen-
sions within each model. The popular model named SERVQUAL consists
of factors like reliability, responsiveness, assurance, empathy and tangibil-
ity. The sub-parts of the eTailQ experience include web design, fulfllment/
authenticity, security/seclusion, and real-time customer support [5].
Performance, fulfllment, device availability, and privacy are all sub-
dimensions of the E-S-QUAL model. Sensual/intuitive, non-cognitive,
132 ◾ Machine Learning for Business Analytics

physical/behavioral, social, and cognitive variables all play a role in the

consumption experience model. Sensorial, sentimental, rational, realistic,
standard of living, and associative components all together build up the cus-
tomer experience model [5].
Another level of complexity of the social climate, interface of the service,
retail atmosphere, variety, price, interactions of consumers in alternative
channels, and retail brand are all parts of the customer experience concep-
tual model. Product experience, outcome concentration, moments of reality,
and peace of mind make up the fnal component of the EXQ system [5].

11.2.3 The Customer Experience Measurement

Although theorizing about customer experience has progressed signifcantly,
measuring customer experience in service research has received little atten-
tion (Lipkin, 2016). In service analysis, the terms customer experience and
service experience are often used interchangeably, but they are not synony-
mous (Klaus and Maklan, 2012) [6].
To evaluate customer experience in retail sector, two approaches are sug-
gested. First approach measures customer experience based on customer’s
perceptions or evaluative judgements of elements leading to service experi-
ence. Second approach measures customer experience based on structural
equation modeling developed [6].
Machine learning (ML) algorithm is an important tool to understand the
customer behavior. During buying process, customer might come across
various advertisements and promotions for which identifying perceptions
and human behavior is sometimes more complex to quantify. Sometimes,
we couldn’t fully understand how the algorithm works for these kinds of
scenarios [7].

11.3 Artifcial Intelligence in the Public Sector

The potential applications of AI are vastly increasing with day-to-day
requirements of consumers involved across the globe. Among them, public
sector benefts from AI through several ways like public safety, social wel-
fare, and public health [1].
Using AI in public safety helps in reducing time delays of patients being
admitted. Also, prioritizing patients who need ambulance and who needed
to treat on site [1].
 Impact of AI on Buyer Experience ◾ 133

Using AI in social welfare mainly advises to public employees to resolve

complex issues arise frequently. And fnally using AI in public health is to
identify the patients having similar symptoms in different locations and
spread awareness of the disease [1].
When the three aspects are applied to the public sector, it is possible to
provide better service to all users while still increasing national security and
satisfaction.

11.4 Artifcial Intelligence in Retail Sector

The retail industry is undergoing a signifcant transformation. A business as
a whole is attempting to keep up with rapidly changing consumer data and
to have suffcient value to transform to online mode rather than traditional
mode [8].
In retail sector, AI-controlled brilliant mirror utilized stores for making vir-
tual perception of garments to the customer as an experience without wear-
ing and letting how it looks bringing new age environment throughout [8].
The key technologies involved for above mentioned smart mirrors have
RFID controlled garment racks, gyro-sensors, and function low vitality chips.
These mirrors are also useful for beauticians to change lighting of the mirror
as per the mood set of customers and make customer delight while provid-
ing services accordingly. These technologies will enhance for future visit of
customers to the store [8].

11.5 Artifcial Intelligence in B2B E-commerce

The growth of e-commerce has increased the value of SCM as businesses
reengineer processes as they migrate to the internet.
A range of AI-based SCM problem-solving methods are available. Most
of them are agent-based, with each agent in charge of one or more SCM
operations. Most agent-based e-commerce approaches concentrate on trans-
actional information, that is, the knowledge needed to negotiate on price,
delivery date, product quantity, and so on [9].
Some agent-based methods will deal with knowledge that isn’t transac-
tional, such as SCM knowledge. Humans or organizations can be modeled
using certain agent-based approaches to solving SCM problems. The models
are used for risk value analysis and simulation (Swaminathan et al. 1998) [9].
134 ◾ Machine Learning for Business Analytics

The foremost goal of AI is to build and create a system that is capable of

imitating the human behavior. An e-commerce system with AI components
can function more “naturally” to its users [9].
Key dimensions that play a major role in deciding and evaluating the AI
applications to e-commerce include naturalness, performance, and creating a
better user interface that facilitates interactions [9, 10].

11.6 Challenges of Artifcial Intelligence

Even with better enhancements of AI happening day to day, but still they
are situations where machines or processes going out of control leading to
effect human race. These constitute challenges of AI in customer experience
and customer buying journey. The major challenges identifed in the discus-
sion with examples are as follows.
Microsoft used AI to build a bot that can have automated conversations
with Twitter users while mimicking their language. The bot has become
offensive and has been labeled racist in less than 24 hours [1].
Google has apologized for a bug in Google Photos that caused images
of black people to be mislabeled as gorillas. A sizable portion of customers
believe that using AI in investments is riskier than it is advantageous [1].
When it comes to the complexities of AI in marketing, the most challenging
part of AI integration is technological compatibility. Person named Waqar Haider
says the company is successful in integrating their system with major CRM sys-
tems to solve the issue of compatibility. It remains a signifcant obstacle for us,
and the organization is constantly working to strengthen the operation [11].
Coming to the challenges for AI in healthcare, there are various factors
identifed. The main challenges include the need for specifc architecture in
businesses, public perceptions of AI, the need for privacy and data security,
and the need for high reliability and service quality [12].

11.7 Benefts of a Satisfed Customer

Customer satisfaction is important because it encourages consumers to feel a
sense of togetherness, sentimental connection, and allegiance of the brand.
Customer satisfaction is defned as “whether a product or service meets a
customer’s needs or demands” or “whether a customer believes he receives
 Impact of AI on Buyer Experience ◾ 135

benefts he always wanted from the goods or services for which the amount
has been paid to a specifed frm.” [13].
According to a survey that is held, the following observations are drawn:

86% of adults spend more on the product of a particular brand

they feel loyal about. If their shopping experience is positive, more
than 70% of people tell other shoppers about it. 43% of the adults
purchase so called inferior products if their experience with the
product is positive [1].
The above three fndings demonstrate that providing a superior
customer experience is critical for increasing revenue, customer
loyalty, and customer base. Delivering superior service creates a
competitive edge [1].

11.8 Research Methodology

Regarding the methodology, a qualitative study on Palestine-based com-
panies along with CEO and IT specialists are chosen for the analysis and
through that later quantitative fndings are identifed in the discussion [1].
For analyzing consumer buying journey and purchase process, applica-
tions of AI like chatbots are induced to know the levels of performance in
purchase process [14].

11.8.1 Population
Based on intense literature review, as mentioned earlier, technology-based com-
panies are considered where CEO and IT specialists along with internet uses
residing in Palestine West Bank are taken as population for the research [1].

11.8.2 Sample Size

Two companies are considered for investigation whether they can go for AI
implementation [1].

11.8.3 Data Sources

Primary Data: Firsthand data collected based on research.
136 ◾ Machine Learning for Business Analytics

Interviews: Interview with CTO of online booking company named

“YAMSAFER.” Interview with a CEO of Palestinian banking company
names The National Bank (TNB) [1].
Questionnaire: The closed-ended questionnaire was distributed to investi-
gate internet users’ actions on AI [1].
Secondary data: Obtaining data that has already been analyzed by other
researchers [1].

Now for analyzing consumer buying journey using chatbots, Krippendorff

(2004) [15] is a distinct technique used when researchers try to avoid recep-
tive situations on two grounds: misrepresenting the fndings, undermining
the study’s validity, and data manipulation by the thesis or origins being
studied [14].
Information is obtained through systematic and face-to-face observation,
wherein ten retailers are being observed conducting their retailing opera-
tions in the Romanian online market in the position of a “mysterious cus-
tomer.” [14].
The content review concentrated on Romanian retailers with the highest
number of consumers. Because of the growth of electronic commerce and
the potential for multiplying e-commerce transactions established in accor-
dance with the Western model, we have considered the Romanian market
for the analysis [14].
The websites were selected based on the number of people who visited
online stores in August 2020. The criteria for selection are supported with a
fact that nearly 70% of Romania’s 199.4 lakh citizens spend time surfng the
internet.

11.8.4 Conceptual Model

Figure 11.1 depicts the related variables in this analysis that the authors used
to evaluate the study hypothesis to emphasize the relationship between
AI and customer experience. The independent variable is AI, whereas
the dependent variable is customer experience, which is divided into two
dimensions: personalized customer service and after-sales customer support.
The consumer purchasing journey model will show how simple chatbots
are used, their consistency and effectiveness, response time, and the value
of customer responses. The information was coded (Table 11.1) so that the
results could be analyzed consistently.
 Impact of AI on Buyer Experience ◾ 137

Figure 11.1 Study hypothesis.

To interpret the data, the results of horizontal analysis (to calculate the
number of occurrences of the code) and vertical analysis are being used.

11.9 Discussion of the Findings

For research hypotheses in this section, as well as the corresponding discus-
sion that goes among each hypothesis, both qualitative and quantitative data
fndings are included in this section.

H0–1:There exists no signifcant relationship between AI and customer

experience.
H0–2: There exists no signifcant infuence of AI on customer experience.

According to Table 11.2, the authors of the research paper applied correla-
tion to analyze the signifcance of relationship between AI and customer
experience; fndings indicate that the two variables have a direct and moder-
ately positive relationship [1].
138 ◾ Machine Learning for Business Analytics

Table 11.1 Responses Coded in Content Analysis

Topic Coding
Using Chatbots 1. Dont use any chatbots
2. lnstant replies
3. Chatbot in Facebook Messenger (full set
version)
Response time 1. Not publicly visible
2. Very prompt
3. In a few hours
4. Within a day
5. More than a day.
Instant message Quality
– Information on 1. Yes
personalized 2. No
recommendations
– Product cost information 1. Yes
2. No
– Information on 1. Yes
personalized assistance 2. No
– Information on popular 1. Yes
personalized products 2. No
– Information on the 1. Yes
availability of products in 2. No
stock
– Using a custom addressing 1. First name only
formula 2. Last name only
3. First name and last name
4. No frst name, no last name
Relevance of replies 1. High
2. Medium
3. Low
Performance 1. Basic sending of automatic messages
2. Proactively recommend content
 Impact of AI on Buyer Experience ◾ 139

Table 11.2 Coeffcient of Correlation between Artifcial Intelligence and

Customer Experience

Dimensions 1 2
Artifcial Intelligence 1 0.514**
Customer Experience 0.514** 1

Table 11.3 ANOVA Table for the Econometric Model of Customer Experience

Model Sum of df Mean f Sig. R R-Square

squares square
Regression 13.571 1 13.571 22.669 0.000
Residual 18.832 89 0.306 0.514 0.264
Total 32.403 90

So, hypothesis (H0–1) concludes that there exists a signifcant relationship

between AI and customer experience.
R-square and ANOVA tests are employed to identify the results relating to
the hypothesis; Table 11.3 shows that the F-distribution of 1 and 89 df has a
signifcant value of 22.669. The signifcance of the regression is established
by using the F-test, with a value of p less than 0.05; hence, the results sug-
gest that AI and customer experience have a signifcant relationship [1].
With the help of R-square and ANOVA reports, it has been identifed that
there exists a positive relationship between AI and customer experience. As
R2 = 0.264, AI estimates 26.4% of the difference in customer experience [1].
The effect of AI on the customer experience was explained by the
authors in form of two aspects (personalized customer service and post-sales
customer support), and the R-square and ANOVA summary statistics are
mentioned in the above table [1].
It has been identifed that AI accounts for 22.9% of the variance in per-
sonalized customer service mentioned in Table 11.4, whereas after sales
customer support accounts for 7% in Table 11.5 [1].
Now moving on to the results and discussion of consumer buying journey
using chatbots, the following content analysis table is explained.
The results of the study paint a picture of support services in the pre-
purchase stage of the customer’s journey, where the customer intends to
connect with the provider by employing Facebook Messenger. By using
140 ◾ Machine Learning for Business Analytics

Table 11.4 ANOVA Table for an Econometric Model of Personalized Customer

Experience

Model Sum of df Mean f Sig. R R-Square

Squares Square

Regression 9.93 1 9.93 24.669 0.000

Residual 33.409 89 0.403 0.479 0.229
Total 43.420 90

Table 11.5 Post-Sales Customer Support (ANOVA TABLE)

Model Sum of df Mean f Sig. R R-Square

Squares Square
Regression 3.641 1 3.641 6.294 0.014
Residual 48.014 89 0.578 0.265 0.07
Total 51.781 90

horizontal and vertical analyses, it has been found that 100% of the major
retailers (Table 11.6) are not using the entire full version of the chatbot [14].
The effect of chatbot content on fnal customer purchasing behavior was
explored by looking at the quality aspects from the perception of the com-
munication started. The following description of each group integrated in the
content analysis was used to present the fndings as concisely as possible.
Although no complete version of a chatbot has been introduced in any of
the ten online stores studied (Figure 11.2), the existence of a particular form
of chatbot in all ten cases analyzed defnes the scenario. This is refected in
the automatic alerts that retailers have set up, as well as the interest in using
AI technology to initiate online experiences with consumers [14].
The supply of in-stock items, the price of customizable products, special-
ized common products, personalized reviews, personalized assistance, and
details on the use of the frst name were all considered while evaluating the
quality of the chatbot [14].
The content analysis coded the performance in two ways: “sending sim-
ple messages” and “recommending content.” The condition in which material
is suggested to the user was not encountered in any of the ten cases exam-
ined. Using the related “simple chatbots,” all of the prominent retailers sent
general conversations, which defned the results obtained [14].
 Impact of AI on Buyer Experience ◾ 141

Table 11.6 Online Store Ranking (Romania)

Place Website Users Visits Views

1 www.pretzmic.ro 51,904 72,308 269,250
2 www.baterii.lux.ro 43,128 59,857 158,099
3 www.uscatorrufe.ro 25,992 31,429 67,343
4 www.wainertools.ro 25,958 31,763 69,993
5 www.magazin-unelte.ro 19,658 24,002 90,991
6 www.vintagetime.ro 19,407 23,954 64,382
7 www.parfumas.ro 12,367 25,350 248,482
3 www.fabricdaelenjerii.ro 11,874 18,720 76,668
9 www.lexservice.ro 11,774 15,343 86,643
10 www.gamestore.ro 11,259 16,279 65,688

Figure 11.2 Online store ranking (Romania) based on the number of consumers.

11.10 Conclusion and Future Work

Using theoretical and conceptual context, the infuence on customer expe-
rience and consumer buying journey is explained with separated analy-
sis identifed. The infuence on customer experience and future work is
explained as follows.
142 ◾ Machine Learning for Business Analytics

The results of the correlational and regression studies show that AI and
customer experience have a good relationship. AI and delivering customized
customer service and after-sales customer care are inextricably linked [1].
Personalized customer service has a positive effect on customer experi-
ence throughout the buying journey of the customer. Customers can wait
less and have a better experience through the use of AI in customer contact
centers and in other areas of service [1].
The study recommended to enhance the customer experience in the
awareness stage. It also advised companies to offer personalized to the cus-
tomer to enhance their experience. According to the report, AI can be used
in call centers and other after-sales support facilities to reduce customer wait
times [1].
Now, the infuence of AI tool chatbots on consumer buying journey and
recommendations are explained as follows.
Consumer frustration is fuelled by the poor quality of content displayed
by simple chatbots, particularly during the pre-purchase stage, when users
want to communicate with retailers in real time [14].
According to the authors, AI can help boost personalized support fea-
tures, even if it is introduced at a simple level, if chatbots are designed to
be capable of engaging in a conversation for a long time on the request of
users, and if the time of response is publicly made available in all of the
Facebook pages of the store retailers, as a result where users loyalty will
magnify [14].
It can be concluded that the decisions made by managers are expressed
in actions developed by retailers in their business functions [14].
The research fnders suggest companies in the global world to keep track
of the levels at which the target public is at and to assess the degree of
acceptance of emerging technology in the context of the courses that poten-
tial clients are taking [14].

References
[1] Daqar, M. A. A., & Smoudy, A. K. (2019). The Role of Artifcial Intelligence on
Enhancing Customer Experience. International Review of Management and
Marketing, 9(4), 22.
[2] Hoyer, W. D., Kroschke, M., Schmitt, B., Kraume, K., & Shankar, V. (2020).
Transforming the Customer Experience through New Technologies. Journal of
Interactive Marketing, 51, 57–71.
 Impact of AI on Buyer Experience ◾ 143

[3] Negroponte, N. (1995). The Digital Revolution: Reasons for Optimism. The
Futurist, 29(6), 68.
[4] Lemke, F., Clark, M., & Wilson, H. (2006). What Makes a Great Customer
Experience?
[5] Havíř, D. (2017). A Comparison of the Approaches to Customer Experience
Analysis. Economics & Business, 31(1).
[6] Meyer, C., & Schwager, A. (2007). Measuring Customer Experience in Physical
Retail Environments. Harvard Business Review, 85, 116–126.
[7] Kaczmarek, J., & Ryżko, D. (2009). Quantifying and Optimizing User
Experience: Adapting AI Methodologies for Customer Experience
Management.
[8] Pabalkar, V., & Nimbalkar, K. (2021). Impact of AI on Retail Sector in India.
European Journal of Molecular & Clinical Medicine, 7(11), 4372–4386.
[9] Sun, Z., & Finnie, G. R. (2004). Intelligent Techniques in E-Commerce. Berlin:
Springer.
[10] Bătăgan, L., Mărăşescu, A., & Pocovnicu, A. (2010). Consumer Rights in Digital
Economy: Case Study of Romanian E-Commerce Usage. Theoretical and
Applied Economics, 9(550), 79–96.
[11] Shahid, M. Z., & Li, G. (2019). Impact of Artifcial Intelligence in Marketing:
A Perspective of Marketing Professionals of Pakistan. Global Journal of
Management and Business Research, 19(2), 1–8.
[12] Iliashenko, O., Bikkulova, Z., & Dubgorn, A. (2019). Opportunities and
Challenges of Artifcial Intelligence in Healthcare. In E3S Web of Conferences
(Vol. 110, p. 02028). EDP Sciences.
[13] Hanif, M., Hafeez, S., & Riaz, A. (2010). Factors Affecting Customer Satisfaction.
International Research Journal of Finance and Economics, 60(1), 44–52.
[14] Nichifor, E., Trifan, A., & Nechifor, E. M. (2021). Artifcial Intelligence in
Electronic Commerce: Basic Chatbots and the Consumer Journey. Amfteatru
Economic, 23(56), 87–101.
[15] Krippendorff, K. (2018). Content Analysis: An Introduction to Its Methodology.
UK: Sage Publications, London.
Chapter 12

Application of Artifcial
Intelligence in Banking:
A Review
Syed Hasan Jafar, Viplav Dhandhukia, and Bijay Kumar G.

Contents
12.1 Introduction............................................................................................ 145
12.2 Literature Review ................................................................................... 146
12.3 Conclusion..............................................................................................146
References....................................................................................................... 151

12.1 Introduction
The use of artifcial intelligence (AI) in banking improves effciency of
banks, saves a lot of time, contributes to cost saving and fnally enhances
customer experience and gives them a better value proposition in terms
of better service, accessibility of various products, and related fnancial
information.
There has been a rise in the use of AI-based tools in banking industry
in recent years. AI enables banks to automate some of its activities and also
helps them in better decision-making with the analysis of a lot of customer
data.
With the implementation of AI, customers get a personalized service,
quicker resolution, or response of their issue or requests which enhances

DOI: 10.4324/9781003206316-12 145

146 ◾ Machine Learning for Business Analytics

overall customer experience; a more customized offering and an overall bet-

ter value proposition. For banks, it saves a lot of time and costs, increased
effciency through automation, enables them to better understand customer
behavior, their needs, and preferences, with which they could come with
better offerings and pursue any new opportunities; serve customers quickly
and with a better personalized customer interaction.
Apart from this, primarily, banking customers require safety and secu-
rity while transacting. With the increased online transactions, the fraudulent
activities associated with online transactions have also increased. The banks
have to bring down this risk and threats associated with online transactions.
Otherwise, customers would not feel secure in carrying on their payment
activities and transactions. So, AI-based tools are being used to for fraud
detection. AI analyzes a large amount of data and understands spending pat-
terns and customer behavior. If there is any odd pattern or unusual behavior
like transacting from a new country, transactions at high frequency or any
other unusual transactions then, AI detects and prevents such a transaction
to take place. Hence, AI is also playing a key role in ensuring security which
enables customers to safely carry on their fnancial transactions.

12.2 Literature Review

The methodology that is used to analyze the different research paper is the
systematic literature review (Table 12.1). These papers show the importance
of AI in banks operations.

12.3 Conclusion
We can infer that AI is advancing quicker than ever in the banking industry,
which is leading to the transformation of lifestyles. Technology and creativity
are flling the void left by the conventional banking system. They work in a
variety of areas of banking, including core banking, customer service, organiza-
tional effciency, and analytics, all of which contribute to better banking experi-
ence. AI aids in the speed of segregation. Initially, the pressure on experienced
employees and remark AI as a threat, but after a period of adapting, they ben-
eftted from new technology and contributed to improve job performance.
AI software will identify solutions for cyber-attack frauds and data
breaches. However, the use of AI-powered technologies is still in its infancy,
Table 12.1 Systematic Literature Review
Paper Objective Finding
Kaur et al. (2020) Improving functional effectiveness of banks AI-powered mobile banking apps
Use of robotic tools in banking processes Use of AI-based algorithms to spot any suspicious activity
and detect frauds.
AI will help in risk management in banking Chatbots and virtual assistants to enhance customer
including operational risk, credit risk, and experience by providing prompt solutions.
market risk
AI will be helpful in fnding credit worthiness Banks can incorporate bank stations which are a system
of clients of AI-based self-service terminals that provide a wide
variety of banking services.
Minimizing banking frauds and scams
Self-service in bank branches
Effective decision-making through the use of
AI tools
Indriasari, Gaol, To explore the application of AI to enhance AI can be used to understand customer behavior,

Application of AI in Banking
and Matsuo the experience of bank customers patterns, and expectations. With this, banks can better
(2019) cater to their customers with improvements in their
product offerings and services. Also, it enables them to
pursue new opportunities.
Current challenges in application of AI in Better value proposition to customers by coming up with
customer service customized offerings after properly understanding their
needs and preferences.
Managers’ perception of implementation of Faster service response can be given after analyzing
AI to enhance customer experience customer behavior and service usage pattern.
Proposing innovative AI-based solutions that AI tools enable customers to have a personalized service

◾
improve banking operations and enhance experience. These tools can understand customer

147
customer experience preferences and can effectively manage their fnances
and goals.
(Continued)
Table 12.1 (Continued)

148 ◾
Paper Objective Finding
Kaya et al. (2019) Application of AI in customer-focused AI is being used for KYC process. AI tools evaluate the

Machine Learning for Business Analytics

applications to enhance customer reliability of information provided by the customers and
experience verify their identity.
Application of AI in banking operations AI algorithms check any suspicious activity in online
transactions and prevent any fraudulent activity.
Use of AI-based tools in fnancial planning The use of chatbots and virtual assistants to address any
and investment advisory services issues or requests of customers quickly without the
involvement of a bank employee.
Use of AI in wealth management for various AI tools to give fnancial solutions and advisory after
set of clients considering factors like customers’ goals and risk-taking
capability.
Improving effciency of banks by automating
certain statutory compliances
Use of AI to prevent and detect any frauds in
online transactions
Ryzhkova et al. Understanding customers’ perception of Some customers have faced diffculties due to technical
(2020) implementation of AI in the banking processes failures and errors in the AI-based system.
Areas where AI can be fully implemented AI tools may sometimes prove to be ineffective. Hence,
after considering current AI tools and some customers are not ready to adopt.
customers’ readiness to adapt
Some customers reacted positively as it improves service
quality in terms of getting quicker solutions and getting
tasks done.
It increases the speed of services. Automation reduces
the time of routine operations and increases effciency.
Overall, most of them are positive with the
implementation of AI in banking.
Paper Objective Finding
Kochhar, Purohit, To get the knowledge of AI application in the AI is assuming an extremely essential part in fnancial
and Chutani banking sector area in lessening cost, alleviating hazard, distinguishing
(2019) misrepresentation, and expanding consumer loyalty,
however then again numerous insurances ought to be
taken while executing it, as there is consistently a danger
of spillage of information which could bring about an
immense misfortune to the banks.
To know the improvisation of banking sector AI in banking—for the individuals who embrace it, AI will
business results with the use of AI over the long haul give a superior encounter to clients
and representatives while conveying genuine business
esteem on each measurement.
AI implementation in the banking sector
Analyzing the impact on Indian Banks with AI
Donepudi (2017) The artifcial intelligence and machine We can pertinently say that banking sector is taking over
learning applications and evaluation in terms by AI and machine learning.
of utility in different functional areas of

Application of AI in Banking
banking
Framework of institutions effciency in terms AI and ML have given the banking sector a new way of
of computational intelligence for meeting their customers’ demands, who are looking for
improvisation of business smarter, convenient, and safer ways to access, save,
spend, and invest their money.
Malali and Reason behind the Indian and fnancial Fintech organizations in the nation are flling in the hole
Gopalakrishnan sectors to go for AI-enabled technologies. gave up by the conventional fnancial area, and this is
(2020) occurring by utilizing the force of innovation and
advancement in artifcial intelligence.
Areas where execution has to be done by AI The clients being emphatically affected by these loaning

◾
technologies. new businesses were never viewed as workable by the

149
conventional fnancial area.

(Continued)
Table 12.1 (Continued)

150
Paper Objective Finding

◾
AI has become the de facto technology used by all

Machine Learning for Business Analytics

fnancial and technology companies to fgure their
platforms.
It is assessed that inside the following decade, AI-fuelled
monetary administrations will be the solitary mechanism
of connection for the clients, making monetary items and
loaning accessible to masses even in the distant towns of
the nation, in this way making monetary incorporation a
nearby reality.
Soni (2019) Cyber security in banks Artifcial intelligence was introduced as a concept to
mimic human brains.
AI advantages and disadvantages in banking AI technology is used for enhancement of customer
communication and experience.
AI for customer interaction Enhancement of the effciency in banking processes
which interns develop the security and controls risk.
Information about cyber-attack and the prices of
solutions is provided by AI toward banks and fnancial
institutions.
Various issues related to fraud and data breach are
identifed by artifcial intelligence techniques.
 Application of AI in Banking ◾ 151

the way they work leads to the growth and advancement in the fnancial
and banking sector, it is possible that the future of AI-powered technologies
will result in minor losses and enhance trade with utmost.

References
Donepudi, Praveen Kumar. “Machine Learning and Artifcial Intelligence in
Banking.” Engineering International 5.2 (2017): pp. 83–86.
Indriasari, Elisa, Ford Lumban Gaol, and Tokuro Matsuo. “Digital Banking
Transformation: Application of Artifcial Intelligence and Big Data Analytics for
Leveraging Customer Experience in the Indonesia Banking Sector.” 2019 8th
International Congress on Advanced Applied Informatics (IIAI-AAI) (pp. 863–
868). IEEE, 2019.
Kaur, Dr, et al. “Banking 4.0: ‘The Infuence of Artifcial Intelligence on the
Banking Industry & How AI Is Changing the Face of Modern-Day Banks’.”
International Journal of Management 11.6 (2020).
Kaya, Orçun, et al. “Artifcial Intelligence in Banking.” Artifcial Intelligence (2019).
Kochhar, Khyati, Harsh Purohit, and Ravisha Chutani. “The Rise of Artifcial
Intelligence in Banking Sector.” The 5th International Conference on
Educational Research and Practice (ICERP) (p. 127). 2019.
Malali, Anil B., and S. Gopalakrishnan. “Application of Artifcial Intelligence and
Its Powered Technologies in the Indian Banking and Financial Industry:
An Overview.” IOSR Journal of Humanities and Social Science 25.4 (2020):
pp. 55–60.
Ryzhkova, Marina, et al. “Consumers’ Perception of Artifcial Intelligence in
Banking Sector.” SHS Web of Conferences. Vol. 80. EDP Sciences, 2020.
Soni, Vishal Dinesh Kumar. “Role of Artifcial Intelligence in Combating Cyber
Threats in Banking.” International Engineering Journal for Research &
Development 4.1 (2019): p. 7.
Chapter 13

Digital Ethics: Toward

a Socially Preferable
Development of AI Systems
C. Guzmán-Velásquez and J. G. Lalinde-Pulido

Contents
13.1 Introduction ............................................................................................ 154
13.1.1 Introduction ................................................................................ 154
13.2 The Role of Data and Algorithms in Today’s Digital Societies............. 155
13.2.1 Data in Digital Economies ......................................................... 155
13.2.2 Artifcial Intelligence, a Look Beyond the Technical
Paradigm ..................................................................................... 156
13.3 Digital Ethics: Its Nature and Scope ...................................................... 158
13.3.1 Ethics of Artifcial Intelligence ...................................................160
13.3.1.1 Ethics as a Service: Ethical Orientation for the
Design of Socially Preferable AI.................................. 161
13.3.1.2 Ethics-Based Auditing as a Governance
Mechanism for Artifcial Intelligence .......................... 161
13.4 Conclusions ............................................................................................ 162
References ....................................................................................................... 163

DOI: 10.4324/9781003206316-13 153

154 ◾ Machine Learning for Business Analytics

13.1 Introduction
13.1.1 Introduction
Every scientifc discipline generates a discourse that enables it to face the
challenges of the area. Biology, geology, and psychology, among many oth-
ers, have their concepts, theories, and methods, which allow researching
and understanding the phenomena they deal with. The specialization of dis-
courses and disciplines has led to the fragmentation of knowledge and the
lack of an understanding of the inter-relationships between different scien-
tifc disciplines (Morley et al., 2019). But history increasingly reminds us that
a fragmented style of thinking is not the most appropriate way to approach
the challenges of this century, defned by the digital revolution (Veliz, 2021).
The digital revolution has had a special impact on the way we understand
certain essential concepts for life, such as time, space, and human nature
(Floridi, 2007). This means that the conceptual tools that we brought are no
longer enough to give an explanation and meaning to the new dynamics
and digital realities (Taddeo & Floridi, 2018). Therefore, science and philoso-
phy acquire a new task of designing concepts to encompass the new envi-
ronments the information societies imply (Floridi, 2015). Therefore, aimed
the so-called digital revolution, and with all the chaos that it can cause when
trying to understand its effects, this article chapter to do an understanding
exercise, facing a problem that is gradually affecting each of the spheres of
human life and scientifc disciplines: the lack of bridges between engineer-
ing, science, and ethics (De Cremer & Kasparov, 2021)
Current times are characterized by how information and communication
technologies are no longer only related to our individual and social well-
being; our well-being and our societies are supported by these technologies
(Floridi, 2021). Thanks to them, and especially to the internet, information
societies have been able to establish themselves and transcend the physical
and geographic space, bringing unprecedented challenges (Floridi, 2021).
It is due to the digital that new frontiers of interaction and social organiza-
tion are opening, including artifcial intelligence (AI) agents, which leads us
to rethink our role as human beings in the digital environment we inhabit
(Taddeo & Floridi, 2018; Burr et al., 2018).
As humanities and computer science begin to open spaces for conversa-
tion to explore new ways of observing, intervening, and modifying human
behaviors and the contexts they inhabit, digital ethics and social data sciences
emerge. At the same time, philosophy and behavioral sciences seek to advance
 Digital Ethics: Development of AI Systems ◾ 155

in the task of conceptually clarifying the phenomena of digital life and eventu-
ally offer conceptual tools to intervene in behaviors and design contexts.
These theoretical and conceptual developments give a glimpse of the real
impact that technologies such as AI have on the different levels of society.
Having as context a highly digitized society, in which there are more digi-
tal devices connected to the network than human beings, in which offine
life merges with online life, and in which human beings interact more and
more with AI systems without even realizing it (Floridi, 2015), it is necessary
to clarify in greater detail the pillars of digital societies to design a better
Infoesphere (Floridi, 2021). To this end, Section 13.2 identifes the role of
data and algorithms in the functioning of digital societies. Data is the raw
material from which products and services are developed. Algorithms are
relevant in this context because they provide the technology needed to man-
age and interpret the data and thus achieve valuable information. And with
these premises, a more comprehensive defnition of what AI includes: tech-
nical approaches, social practices, and industrial infrastructures (Tsamados
et al., 2020). In Section 13.3, digital ethics is addressed as a particularly rel-
evant conceptual tool for interpreting the problems and challenges of digital
societies, where the use of data and algorithms affects the well-being of the
society and generates problems on a global scale (Öhman & Watson, 2019).
Finally, in Section 13.4, conclusions are presented.

13.2 The Role of Data and Algorithms

in Today’s Digital Societies
The pillars of information societies are data and algorithms. Organizations
and societies use them to make “data-based decisions,” seeking better rel-
evance, precision, and even anticipation in their actions (Provost, 2013).
Thanks to this infrastructure new companies, or apps, can emerge, offering
convenient solutions to their users. Uber, Google, Apple, Amazon, Rappi, and
Facebook, among many others, are examples of such. Data, digital econo-
mies, AI, and some of the ethical aspects that they entail are presented next.

13.2.1 Data in Digital Economies

Data is defned as: “a symbolic representation, be it numerical, alphabetic,
spatial, etc., of a quantitative or qualitative attribute or variable” (Floridi
156 ◾ Machine Learning for Business Analytics

et al., 2019). Thus, data are partial descriptions of empirical facts, events, and
entities. When a set of data is framed in the light of a hypothesis, theory, or
question, it can be converted into information of value.
It is here where the so-called surveillance economy proposed by
Shoshana Zuboff (2020) begins to work. In current digital societies, technol-
ogy companies collect an abysmal amount of private data about their users,
both individuals and groups, giving them unprecedented power. It is a huge
concentration of power, as they intensively collect data on people, popula-
tions, and network dynamics, all over the world, without any regulation
(Taddeo & Floridi, 2018; Veliz, 2021). It is necessary to clarify the collection
of data, at any level, since, in most cases, these companies are not interested
in analyzing the data at the individual level; but at the group level, that is
individuals who share characteristics, to infuence their processes decision-
making (Burr & Cristianini, 2019; Floridi, 2021).
The goal is to infuence user decisions, with a more serious implication in
the case of “social networks” such as Facebook and Google, as they sell the
insights they have generated from the vast amount of private data of their
users to the highest bidder. This activity is eroding democratic structures, in
addition to the autonomy of its users (Van Bavel et al., 2020) by fragmenting
the shared narratives that unite societies. When talking about data, it is in
the broad spectrum of the concept. That is, data generated by devices, vehi-
cles, organizations, cities, and climate monitors, among many other sources,
which, thanks to sophisticated measurement sensors, can be collected and
subsequently analyzed to make more precise decisions (Öhman & Watson,
2019). Thus, the data can be used to describe phenomena, to predict pos-
sible outcomes, and to prescribe courses of action.
Having a little more clarity regarding data, let’s move on to a compre-
hensive defnition of AI, the role of algorithms in this digital infrastructure,
and how organizations and societies make use of it of so-called AI to ana-
lyze and make sense of the immense amount of data (Big Data) they col-
lect and use their fndings to make decisions (Mökander & Floridi, 2021;
Mökander et al., 2021).

13.2.2 Artifcial Intelligence, a Look Beyond

the Technical Paradigm
The last 15 years have brought incredible advances in the development of
AI, but the social implications of these technologies have not been given
suffcient relevance (Cowls et al., 2021; Veliz, 2021). It is a historic change,
 Digital Ethics: Development of AI Systems ◾ 157

where the power and scope of these technologies are outstripping the abil-
ity to understand their impact and how they work (Burr et al., 2018). And
as digital technologies and AI affect more areas of daily life, they pres-
ent social, political, and environmental challenges that were previously
unknown (Sunstein, 2008; Öhman & Watson, 2019; Taddeo &
Floridi, 2018).
Much has been said about AI and how it could come to “dominate” and
even “replace” human beings when general AI is fnally achieved (Burr
et al., 2018), attributing to it an omnipotent scope, and therefore, taking
responsibility for its designers and creators (Watson, 2020). But the evolu-
tion of recent years has revealed how this narrative hides particular inter-
ests, especially of Silicon Valley companies (Watson, 2020; Milligan, 2018),
and does not obey reality. The hype has distracted the attention of public
opinion, academics, regulators, politicians, and designers who should devote
their energy to propose comprehensive ways to articulate these technologies
benefcially with society (Floridi, 2019; Burr, 2020).
The idea of AI will be explored from three perspectives to build a
comprehensive vision (Crawford & Joler, 2018; Floridi, 2019; Mökander &
Floridi, 2021):

◾ Technical approaches: Technical advances such as symbolic logic,

expert systems, and machine learning (ML), including different tech-
niques as deep learning or generative adversarial networks. Using the
term intelligent for systems like these is a trap because they are far
from being intelligent (Crawford & Joler, 2018). These are technologies
that identify patterns, group, optimize, and make predictions on large
amounts of data.
◾ Social practices: People who design these systems and decide which
problems will be solved using these technologies. This power deter-
mines not only the types of technologies, including the types of algo-
rithms and data, but also defnes which populations are most benefted
by these tools and which populations are excluded and even discrimi-
nated against (Crawford & Joler, 2018; Mittelstadt et al., 2016).
◾ Industrial infrastructures: AI is a huge infrastructure that requires
a planetary computer network. The economic and environmen-
tal resources needed to maintain these gigantic infrastructures are
immense. Thus, these infrastructures represent a deep concentration of
power (Crawford & Joler, 2018). While a laborer in Mali mining lithium,
to sell to tech companies, can earn $ 1 a day’s work, Jeff Bezos, the
158 ◾ Machine Learning for Business Analytics

former CEO of Amazon, makes approximately $ 270 million a day, and

they both work for the same industry (Crawford & Joler, 2018; Cowls
et al., 2021).

Identifying the true costs of these technologies seems to be a pertinent and

necessary concern. That is why ethical, social, and environmental implica-
tions must be addressed alongside technical issues, since these technologies
have a global impact and time is not on our side (Floridi, 2021; Crawford,
2018). Therefore, digital ethics is a conceptual framework to comprehensively
observe the new dilemmas that these technologies bring, such as inequity, pri-
vacy (Taddeo & Floridi, 2018; Veliz, 2021), autonomy, personal determination
(Van Bavel et al., 2020), and digital sovereignty (Floridi, 2021), among others.

13.3 Digital Ethics: Its Nature and Scope

Due to the digital transformation of society, changes in relationships, ways of
working, decision-making, social norms, interpersonal, and institutional trust
have deep impact in individual and collective life. Thus, ethical refection
becomes increasingly relevant, especially for those professionals and design-
ers of these technologies because they have a planetary scope (Floridi, 2021;
Floridi, 2020; Mittelstadt, 2019).
Considering that the current revolution is the product of advances in
digital technologies, there is a common agreement in defning current societ-
ies as information societies or digital societies (Ohman and Watson, 2019).
This means that information is an asset of economic, political, and social
interest in each of human actions (Floridi, 2007; Floridi et al., 2019; Floridi
& Cowls, 2019; Floridi & Strait, 2020; Mittelstadt et al., 2016, Mittelstadt,
2019). Therefore, digital ethics is responsible for addressing questions such
as: What characteristics does this information have? What data is stored and
how is it used? How are the digital infrastructures built? How are human
beings more persuasive based on the way information is presented? Who
are the designers of these digital infrastructures? On what ethical and moral
values are these digital infrastructures based?
Awareness about the relevance, effects, and possibilities that information
and digital technologies can have in everyday life is just beginning to gain
strength in our society. The gap between those who take advantage of infor-
mation management and those who do not do so is widening and creating
an enormous digital divide (Floridi, 2021; Mittelstadt et al., 2016). According
 Digital Ethics: Development of AI Systems ◾ 159

to UNESCO, it is necessary to consider digital literacy as one of the essential

skills in which it is necessary to train people for the 21st century (Öhman &
Watson, 2019). Additionally, training professionals with ethical foundations
are needed (Morley et al., 2021).
As Carl Ömhan, from the Oxford Internet Institute (2019), synthesizes:

The digital revolution brings enormous opportunities to improve

public and private life, and our environments, from healthcare to
smart cities to global warming. Unfortunately, these opportunities
come with signifcant ethical challenges. In particular, the extensive
use of more and more data, often personal (Big Data), the increas-
ing dependence on algorithms to analyze it to shape options and
make decisions (including machine learning, artifcial intelligence,
and robotics), and the gradual reduction of human participation
or the supervision of automatic processes, raise urgent questions
about fairness, responsibility, and respect for human rights.

Thus, social preferability should be the guiding principle to achieve a

solid ethical balance for any digital project with an impact on human life.
So, Luciano Floridi (2019) understands digital ethics as the branch of ethics
that studies and evaluates moral problems related to

◾ information and data, including its generation, recording, curation, pro-

cessing, dissemination, exchange, and use;
◾ algorithms, including AI, artifcial agents, ML, and robots; and
◾ the corresponding practices and infrastructures, including responsible
innovation, programming, piracy, professional codes, and standards.

All these aimed to formulate and support morally good solutions. Therefore,
a successful way to identify and evaluate socially preferable projects is to
analyze them based on their results. These are successful insofar as they
help reduce, mitigate, or eradicate social or environmental problems, without
introducing new damages or amplifying existing ones. Cowls et al. (2021) go
on to suggest that an AI-based project is socially preferable if it is designed,
developed, and deployed in a way that (1) prevents, mitigates, and/or solves
problems that negatively affect human life and/or the well-being of the natu-
ral world and/or (2) allows the development of socially desirable or environ-
mentally sustainable activities and in turn (3) does not introduce new forms
of damage and/or amplify existing inequities.
160 ◾ Machine Learning for Business Analytics

13.3.1 Ethics of Artifcial Intelligence

Algorithms have become crucial agents for the entire digital infrastructure
that digital and information societies currently inhabit. Governments, educa-
tional institutions, companies, courts of law, and hospitals depend on AI sys-
tems to make crucial decisions. And while delegation of tasks to AI systems
can improve effciency and enable new solutions, these benefts are accom-
panied by ethical challenges (Mökander et al., 2021). For example, these
technologies can produce discriminatory results, violate individual privacy,
and undermine human self-determination (Veliz, 2021; Zuboff, 2020).
Therefore, new governance and forecasting mechanisms are needed (Floridi
& Strait, 2020) to help organizations to design and implement these technolo-
gies in an ethical manner, while fostering the economic and social benefts
of AI (Floridi & Cowls, 2019; Floridi, 2021). As a result, organizations have
launched initiatives to establish ethical principles for the adoption of socially
preferable AI. Among these initiatives are the Asilomar principles for AI (2017),
the Declaration for Responsible Artifcial Intelligence (Van Bavel et al., 2020),
the principles offered by the IEEE in its article Ethically Aligned Design (Floridi,
2020), the ethical principles offered by the European Group of Ethics in
Robotics (Floridi, 2021), and the fve General Ethical Principles for a code of AI
of the House of Lords of the United Kingdom (Morley et al., 2021). There is a
good overlap among the fve sets of principles, showing that democratic values
are present when human rights serve as a guide (Floridi et al., 2019).
When these sets are reviewed in the light of the four bioethical
principles—benefcence, non-malefcence, autonomy, and justice, the har-
mony between the sets of principles is more clear (Floridi, 2020). But,
according to the Digital Ethics Lab headed by Luciano Foridi, it is necessary
to add a new principle that ensures the intelligibility and responsibility of
the actions carried out by AI systems: explicability (Floridi & Cowls, 2019).
The following are the non-exhaustive descriptions of the principles for AI:

◾ Benefcence is to promote well-being, preserve dignity, and take care of

the planet.
◾ Non-malefcence is doing no harm, particularly concerning the preven-
tion of breaches of privacy, security, and the ability to be cautious.
◾ Autonomy by implementing AI, some decision-making power is ceded
to these technological artifacts. Autonomy means reaching a balance
between the decision-making power kept and that delegated to artifcial
agents.
 Digital Ethics: Development of AI Systems ◾ 161

◾ Justice: AI should contribute to global justice and equitable access to the

benefts of digital technologies. Promoting prosperity and preserving
solidarity.
◾ Explicability is based on two objectives: (1) intelligibility—that is, how
does the system work?—and (2) responsibility, in the ethical sense—
that is, who answers for the way the system works?

13.3.1.1 Ethics as a Service: Ethical Orientation for

the Design of Socially Preferable AI
Ethics as a service propose the mediation of managers and software devel-
opers so that AI systems are build using socially preferable technologies
and solutions. Two tensions must be constantly assessed in the search for
ethically sound AI systems: the tension between too fexible and too strict.
As Crawford & Joler (2018) point out, algorithmic systems are assemblages
between human and non-human agents that have many non-deterministic
impacts, which is why it is necessary to understand the infrastructure where
the AI system operates and its relationship with laws, regulations, and stan-
dards. The other tension arises between delegated responsibility and cen-
tralized responsibility, which refers to when to request an external audit
and what aspects to audit with the external auditor. External ethical audit
must be a central component of any operationalization of AI ethics (Morley
et al., 2021). In addition, conficts of interest in internal audits can lead to
the inability to maintain an objective opinion by the auditor, while external
audits have limitations such as not having access to all relevant information
about the system due to contractual or privacy terms.
In the digital and information societies where the ethics of AI serves as
an infrastructure for the appropriate design of socially preferable AI systems,
a methodology must be available to defne and implement ethical principles
in the systems, complemented with an audit system that allows the iterative
and constant review of its evolution.

13.3.1.2 Ethics-Based Auditing as a Governance

Mechanism for Artifcial Intelligence
There is a signifcant gap between the theory of the ethical principles of AI
and their practical application (Morley et al., 2021). Therefore, it is necessary
to fnd ways to include the principles, defned by ethical codes and frame-
works, in AI systems while constantly reviewing and following governance
162 ◾ Machine Learning for Business Analytics

mechanisms that allow verifying the appropriate materialization of the

principles in AI systems. Thus, ethics-based auditing is proposed as the
governance mechanism that aims to close the gap between the what (ethical
principles) and the how (implementation) of AI ethics, serving as a gover-
nance mechanism for verifcation of the implementation of ethical principles.
Ethics-based auditing does not mechanize ethics but helps to identify,
visualize, and communicate the ethical and moral values that are framed
in an AI system (Morley et al., 2019). The responsibility for identifying and
executing the steps to ensure that AI systems are ethically sound rests on the
management of the organizations that design and deploy them (Cowls et al.,
2021).
In ethics-based auditing, the auditor ensures the correct questions are
addressed and answered appropriately in the design of the AI systems
(Morley et al., 2021). The conversation starts from principles and goes toward
the directive/managerial intervention (implementation) through the product
cycle. Additionally, due to the changing nature of AI systems, the audit exer-
cise must be continuous to fnd out when the system may be causing dam-
age or behaving in unexpected ways (Cowls et al., 2021).

13.4 Conclusions
Actual societies increasingly depend on digital technologies for their opera-
tion. Among the new technologies, AI systems and the data sources neces-
sary for their operation provide great power to their designers, which can
affect from individuals to entire social systems. Therefore, organizational
devices such as ethics as a service are increasingly necessary for the proper
materialization of ethical principles in the design and implementation of AI
systems. Additionally, governance and control systems such as ethics-based
auditing become pillars of the ethical infrastructure (Floridi et al., 2020) in
digital societies, to design socially preferable systems and solutions. This is
a feld that is in the early stages of its development and that still needs time
and experimentation to identify the limitations that devices and practices
such as ethics as a service and ethics-based auditing may have.
These are devices that enable the governance of AI systems by providing
tools for visualization and monitoring their results and seeking to transcend
the black-box model. Also, once implemented internally and externally by
organizations and governments, they will allow citizens and users to be kept
informed of how the decisions that an AI system has reached and to submit
 Digital Ethics: Development of AI Systems ◾ 163

them to new consideration. Not only seeking to alleviate human suffering and
anticipate the possible harm caused by these technologies but also, and more
importantly, assigning the responsibilities to the right people and organizations.

References
Burr, C., & Cristianini, N. (2019). Can Machines Read Our Minds? Minds &
Machines, 29, 461–494. https://doi.org/10.1007/s11023-019-09497-4
Burr, C., Cristianini, N., & Ladyman, J. (2018). An Analysis of the Interaction
between Intelligent Software Agents and Human Users. Minds & Machines, 28,
735–774. https://doi.org/10.1007/s11023-018-9479-0
Burr, C., & Morley, J. (2020). Empowerment or Engagement? Digital Health
Technologies for Mental Healthcare. In: Burr C., Milano S. (eds.) The 2019
Yearbook of the Digital Ethics Lab. Digital Ethics Lab Yearbook. Springer,
Cham. https://doi-org.ezproxy.eaft.edu.co/10.1007/978-3-030-29145-7_5
Cowls, J., Tsamados, A., Taddeo, M. et al. (2021). A Defnition, Benchmark and
Database of AI for Social Good Initiatives. Nat Mach Intell, 3, 111–115. https://
doi.org/10.1038/s42256-021-00296-0
Crawford, Kate, & Joler, Vladan. (2018, September 7). Anatomy of an AI System:
The Amazon Echo As An Anatomical Map of Human Labor, Data and
Planetary Resources. AI Now Institute and Share Lab. Available at: https://
anatomyof.ai
De Cremer, D., & Kasparov, G. (2021). The Ethical AI—Paradox: Why Better
Technology Needs More and Not Less Human Responsibility. AI Ethics.
https://doi.org/10.1007/s43681-021-00075-y
Floridi, L. (2007). A Look into the Future Impact of ICT on Our Lives. SSRN
Electronic Journal. 23(1), 59–64.
Floridi, L. (2015). The Onlife Initiative. The Onlife Manifesto. In: The Onlife
Manifesto. Springer, Cham. https://doi.org/10.1007/978-3-319-04093-6_2
Floridi, L. (2020). What the Near Future of Artifcial Intelligence Could Be. In:
Burr C., Milano S. (eds.) The 2019 Yearbook of the Digital Ethics Lab. Digital
Ethics Lab Yearbook. Springer, Cham. https://doi-org.ezproxy.eaft.edu.
co/10.1007/978-3-030-29145-7_9
Floridi, L. (2021, June 1). The European Legislation on AI: A Brief Analysis of Its
Philosophical Approach. Available at SSRN: https://ssrn.com/abstract=3873273
or http://dx.doi.org/10.2139/ssrn.3873273
Floridi, L., Cath, C., & Taddeo, M. (2019). Digital Ethics: Its Nature and
Scope. In: Öhman C., Watson D. (eds.) The 2018 Yearbook of the Digital
Ethics Lab. Digital Ethics Lab Yearbook. Springer, Cham. https://doi.
org/10.1007/978-3-030-17152-0_2
Floridi, L., & Cowls, J. (2019, September 20). A Unifed Framework of
Five Principles for AI in Society. Available at SSRN: https://ssrn.com/
abstract=3831321 or http://dx.doi.org/10.2139/ssrn.3831321
164 ◾ Machine Learning for Business Analytics

Floridi, L., Cowls, J., King, T.C. et al. (2020). How to Design AI for Social Good:
Seven Essential Factors. Science and Engineering Ethics, 26, 1771–1796.
https://doi.org/10.1007/s11948-020-00213-5
Floridi, L., & Strait, A. (2020). Ethical Foresight Analysis: What It Is and Why
It Is Needed? Minds & Machines, 30, 77–97. https://doi.org/10.1007/
s11023-020-09521-y
Milligan, M. (2018, April 12). Technology and the Ethics Gap. ABET. Available at:
www.abet.org/technology-and-the-ethics-gap/
Mittelstadt, B. (2019, November). Principles Alone Cannot Guarantee Ethical AI
(May 20, 2019). Nature Machine Intelligence. Available at SSRN: https://ssrn.
com/abstract=3391293 or http://dx.doi.org/10.2139/ssrn.3391293
Mittelstadt, B., Allo, P., Taddeo, M., Wachter, S., & Floridi, L. (2016, December).
The Ethics of Algorithms: Mapping the Debate. Big Data & Society.
doi:10.1177/2053951716679679
Mökander, J., & Floridi, L. (2021). Ethics-Based Auditing to Develop
Trustworthy AI. Minds & Machines, 31, 323–327. https://doi.org/10.1007/
s11023-021-09557-8
Mökander, J., Morley, J., Taddeo, M. et al. (2021). Ethics-Based Auditing of
Automated Decision-Making Systems: Nature, Scope, and Limitations. Science
and Engineering Ethics, 27, 44. https://doi.org/10.1007/s11948-021-00319-4
Morley, J., Elhalal, A., Garcia, F. et al. (2021). Ethics as a Service: A Pragmatic
Operationalisation of AI Ethics. Minds & Machines, 31, 239–256. https://doi.
org/10.1007/s11023-021-09563-w
Morley, J., Floridi, L., Kinsey, L., & Elhalal, A. (2019). From What to How: An
Initial Review of Publicly Available AI Ethics Tools, Methods and Research
to Translate Principles into Practices. SSRN Electronic Journal. https://doi.
org/10.2139/ssrn.3830348
Morley, J., Morton, C., Karpathakis, K., Taddeo, M., & Floridi, L. (2021). Towards
a Framework for Evaluating the Safety, Acceptability and Effcacy of AI
Systems for Health: An Initial Synthesis. SSRN Electronic Journal. https://doi.
org/10.2139/ssrn.3826358
Öhman, C., & Watson, D. (2019). Digital Ethics: Goals and Approach. In: Öhman
C., Watson D. (eds.) The 2018 Yearbook of the Digital Ethics Lab. Digital Ethics
Lab Yearbook. Springer, Cham. https://doi.org/10.1007/978-3-030-17152-0_1
Provost, F., & Fawcett, T. (2013). Data Science and Its Relationship to Big Data and
Data-Driven Decision Making. Big Data, 1(1), 51–59.
Sunstein, C. (2008). Democracy and the Internet. In: Van den Hoven J., Weckert
J. (eds.) Information Technology and Moral Philosophy (Cambridge Studies
in Philosophy and Public Policy, pp. 93–110). Cambridge University Press,
Cambridge. doi:10.1017/CBO9780511498725.006
Taddeo, M., & Floridi, L. (2018). How AI Can Be a Force for Good. Science,
361(6404), 751–752. https://doi.org/10.1126/science.aat5991
 Digital Ethics: Development of AI Systems ◾ 165

Tsamados, A., Aggarwal, N., Cowls, J., Morley, J., Roberts, H., Taddeo, M., &
Floridi, L. (2020, July 28). The Ethics of Algorithms: Key Problems and
Solutions. Available at SSRN: https://ssrn.com/abstract=3662302 or http://
dx.doi.org/10.2139/ssrn.3662302
Van Bavel, J. J., Harris, E. A., Pärnamets, P., Rathje, S., Doell, K. C., & Tucker, J. A.
(2020) Political Psychology in the Digital (Mis)Information Age: A Model of
News Belief and Sharing. Journal of Psychological Study of Social Issues, 15,
84–113.
Veliz, C. (2021). Privacy Is Power. Random House, UK.
Watson, D. (2020). The Rhetoric and Reality of Anthropomorphism in Artifcial
Intelligence. In: Burr C., Milano S. (eds.) The 2019 Yearbook of the Digital
Ethics Lab. Digital Ethics Lab Yearbook. Springer, Cham. https://doi-org.
ezproxy.eaft.edu.co/10.1007/978-3-030-29145-7_4
Zuboff, S. (2020). The Age of Surveillance Capitalism: The Fight for a Human
Future at the New Frontier of Power. Public Affairs.
Index

Note: Page numbers in italics indicate a fgure and page numbers in bold indicate a table
on the corresponding page.

A artifcial intelligence, 13, 45–46, 49, 50,

52, 58
actuarial insurance and artifcial in actuarial insurance, 80–84
intelligence, 80–84 adoption in digital marketing, 104–105
Adaptive Neuro-Fuzzy Inference System in agriculture, 88–95
(ANFIS), 20, 24, 25 in agri supply chain, 91
ADT pulse, 79 in B2B e-commerce, 133–134
aerial assessors and artifcial intelligence, in B2B marketing, 107–115
79 and B2B sales funnel, 112, 113
affective customer experience, 122 in banking, 145–151
after-sales customer support, 130, 139, 140 brilliant, 75
agent-based modeling, 19, 24 in business, 34–40, 48–50, 75–77
agriculture in car insurance, 80
and artifcial intelligence, 88–95 challenges of, 134
chatbots, 91 and co-creation of B2B, 112, 112
drones, 91 correlation between customer
future of artifcial intelligence in, 92–93 experience, 117–127, 136–141, 137,
in India, 89–90 138, 139, 140, 141
limitations of artifcial intelligence in, 94 in crop and soil monitoring, 90
precision farming, 92 defnition, 88, 99, 119–120
supply chain and artifcial intelligence, digital ethics and, 153–163
91 ethics-based auditing, 161–162
AI, see artifcial intelligence ethics of, 160–162
algorithms, 2 evolution of, 88–89
ethics and, 160, 161 fragile, 75
reinforcement learning, 16 future in agriculture, 92–93
supervised learning, 15–16 in health insurance, 79
types of, 15–16 in insurance claims, 77–78
unsupervised, 16 in insurance for aerial assessors, 79
Amazon.com, 124 in insurance for self-driving cars, 78, 79
Artifcial General Intelligence (AGI), 75

167
168 ◾ Machine Learning for Business Analytics

intellect levels of, 75, 76 growth and regression models, 37

limitations in agriculture, 94 performance evaluation and regression
management’s perceptions and models, 35
awareness of, 102–104
and market knowledge in B2B, 112–113, C
113
in medical imaging and diagnostics, 78, car insurance and artifcial intelligence, 80
78 cement industry using machine learning,
in person to person insurance, 78–79, 79 22
and precision farming, 92 chart analysis, 77
predictive agriculture analytics, 90 chatbots, 45–55, 100, 109, 124, 125, 130,
in the public sector, 132–133 134, 136, 140
in the retail sector, 133 in agriculture, 91
robust, 75 limitations, 53–54
and smart homes insurance, 79 satisfaction, feedback and awareness on,
Artifcial Narrow Intelligence (ANI), 75 51–53, 52–53
artifcial neural network (ANN), 19, 22, 24 cities, sustainable promoted by machine
auditing, ethics-based, 161–162 learning, 19
Automated Endoscopic System for Optimal Classifcation and Regression Tree (CART),
Positioning (AESOP) system, 60 20, 24, 25
clean energy and machine learning, 19–20
B commercial data processing, 7
Computer Motion, 60
B2B marketing computers, 2, 4
and artifcial intelligence, 107–115 and large-scale data processing, 7
challenges of artifcial intelligence, computer science, 2–3
110–111 conceptual model of customer experience,
co-creation and artifcial intelligence, 131, 132
112, 112 conditional random feld, 100
e-commerce and artifcial intelligence, confusion matrix, 83
133–134 Consumer Behaviors DNA algorithm, 120
sales funnel and artifcial intelligence, consumption experience model, 131–132
112, 113 corporate misconduct, 21
banking and artifcial intelligence, 145–151 correctness of data, 10
Big Data, 3, 4, 5, 103, 159 credit score, 34
and data visualization, 11 crop and soil monitoring with artifcial
defnition, 99 intelligence, 90
binomial logistic regression, 32, 35 curative model of linear regression, 31
bots, see chatbots customer experience
Brazil sugar industry using machine activities of, 124, 124–125
learning, 23 affective, 122
brilliant artifcial intelligence, 75 and artifcial intelligence, 77, 117–127,
business, see also digital marketing; 129–143, 137, 138, 139, 140, 141
marketing cognitive/creative business performance,
artifcial intelligence in, 77 122
failure prediction and regression models, conceptual model of customer
36 experience, 131, 132
 Index ◾ 169

consumption experience model, 131–132 electronic procession, 8

customer experience model, 131 exploratory data analysis, 10–11
defnition, 118–119, 131 group, 7–8
dimensions of, 130, 131–132 large-scale/commercial, 7
E-S-QUAL, 131 real-time, 8
eTailQ Experience, 131 research/scientifc, 7
EXQ framework, 131 types of, 7–8
measurement, 132 data quality, 10
physical customer service (Act), 122 data science, 5
satisfaction factors, 120–121, 121 data scrubbing, 8–10
SERVQUAL model, 131 data transformation, 9
social-identity, 122–123 data visualization, 11
types of, 121–123 data warehouses, 9
customer experience management, 119 DaVinci Surgical System, 60, 62, 64–65
customer experience model, 131 deep neural network, 24
customer personalization, 120, 130, 139, demand forecasting and regression models,
140, 142 39
customer satisfaction, 134–135 descriptive data analysis, 5
cybersecurity applications of machine diagnostic data analysis, 5
learning, 13 digital ethics, 153–163
digital marketing, see also business;
D marketing
and artifcial intelligence, 97–106
data, see also Big Data management’s perceptions of artifcial
authenticity of, 10 intelligence, 102–103
basics, 3–16 digital societies, 155–158
defnition, 3–4, 155–156 direct marketing score model, 36
in digital economies, 155–156 drones
raw, 3 in agriculture, 91
units of measurement, 4 and artifcial intelligence, 79
unstructured, 4
data analysis, 3 E
defnition, 4
descriptive, 5 e-commerce, see also digital marketing
diagnostic, 5 and artifcial intelligence, 133–134
exploratory, 10–11 EDA, see exploratory data analysis
and machine learning, 13 e-government services for citizens and
predictive, 6 machine learning, 21
prescriptive, 6 electronic data processing, 3
types of, 5–6 electronic possession, 8
data cleaning, 8–10 employee commitment, 38
data mining, 6–11 employee selection and regression models,
defnition, 6 38
data pre-processing, 8 energy and machine learning, 19–20
data processing, 3, 6–8 E-S-Qual, 131
data cleaning, 8–10 eTailQ Experience, 131
data pre-processing, 8 ethics, digital, 153–163
170 ◾ Machine Learning for Business Analytics

exactness of data, 10 I
Excelsius GPS, 62
exploratory data analysis, 10–11 India’s agriculture, 89–90
export market and regression models, 37 insurance
EXQ framework, 131 business models, 77
claims and artifcial intelligence, 77–78,
79
F
client risk rate, 82
Facebook ad manager, 103, 104 health, 79
Facebook insights, 102 person to person, 78–79, 79
face recognition, 83, 84 for smart homes, 79
family business research and regression insurance industry
models, 36–37 artifcial intelligence in, 75–86
fnance and regression models, 34–36 issues in, 74
fragile artifcial intelligence, 75 Internet of Things (IoT), 78
fraud detection, 83 Intuitive Surgical, 64, 65, 66
IoT, see Internet of Things (IoT)
G
L
genetic algorithm (GA), 22
geothermal energy and machine learning, large-scale data processing, 7
19–20 lasso regression, 32, 36
Google ad words, 104 Latent Dirichlet Allocation (LDA) algorithm,
Google Analytics, 102 23, 25
Google Data Studio, 102, 104 linear regression models, 31, 38, 40
government applications of machine location planning and machine learning,
learning, 13 19–20, 24
government services and machine learning, logistic regression, 31–32
21 long short-term memory, 23, 25
green customers and machine learning, 20,
25 M
green supply chain management and
regression models, 40 machine learning, 3
group data processing, 7–8 adoption in digital marketing,
104–105
ambiguity and probability in, 14–15
H
applications of, 13–14
healthcare applications of machine assisting green customers, 20
learning, 13 for clean energy, 19–20
Higher-Order SVD (HOSVD) technique, 20, and customer experience, 132
24, 25 and data analysis, 14
hospitality industry, 20, 24 and data mining, 6
chatbots in, 45–55 defnition, 99
human capital development and regression and digital marketing, 97–106
models, 38–39 for e-government services for citizens,
human resource management and 21
regression models, 38–39 impact on customer’s experience, 118
 Index ◾ 171

in insurance industry, 77 P
introduction, 12–16
management’s perceptions and partial least-squares regression model, 35
awareness of, 102–104 payroll and compensation and regression
necessity of, 12 models, 39
for people, 20–21 personalized customer service, 120, 130,
for the planet, 18–20 139, 140, 142
for proft, 21–23 person to person insurance, 78–79, 79
promoting shareholders’ interest, 21 physical customer experience, 122
for promoting sustainable cities, 19 planet and machine learning, 18–20
and sustainability, 17–27, 24 precision farming, 92
for waste management, 19 predictive agriculture analytics with
machine learning algorithms, see artifcial intelligence, 90
algorithms predictive data analysis, 6
Mailchimp, 104 prescriptive data analysis, 6
MAKO, 65 price determinants and regression models,
market consumption and regression 37
models, 37 probability for machine learning, 14–16
marketing, see also business; digital PROBOT, 60
marketing proft and machine learning, 21–23
and regression models, 36–37 public health and artifcial intelligence,
market knowledge in B2B, 112–113, 133
113 public safety and artifcial intelligence,
MAZOR X, 62 132
medical imaging and diagnostics and public sector and artifcial intelligence,
artifcial intelligence, 78 132–133
medical industry and robotics, 57–72 PUMA 560, 60
multi-nominal logistic regression, 32 Python, 3
multiple linear regression model, 31, 35,
37, 39 Q
multi-variate regression, 39, 40
municipal solid waster, see waste quantile regression, 33, 35–36, 37
management
R
N
random forest regression, 33
natural language generation, 49 random forest (RF) model, 21, 24
natural language processing, 49, 77 real-time data processing, 8
natural language understanding, 49 regression models, 29–43
navigational robots, 59 in business applications, 34–40
and business growth, 37
and business performance evaluation,
O
35
online marketing, see digital marketing and demand forecasting, 39
operations and regression models, 39–40 direct marketing score model, 36
ordinal logistic regression, 32, 35 and employee selection, 38
organizational justice, 38 and export market, 37
172 ◾ Machine Learning for Business Analytics

and family business research, 36–37 robots

in fnance, 34–36 navigational, 59
green supply chain management, 40 rehabilitation, 59
in human capital development, 38–39 skin modifcation, 59
in human resource management, 38–39 surgical, 58–60, 61
lasso regression, 32, 36 robust artifcial intelligence, 75
limitations of, 33–34 R programming language, 3
logistic regression, 31–32
market consumption and price S
determinants, 37
in marketing, 36–37 sales and marketing applications of
multiple linear regression, 31, 35, 37, 39 machine learning, 13
in operations, 39–40 scientifc data processing, 7
organizational justice and employee self-driving cars and artifcial intelligence
commitment, 38 for insurance, 78
payroll and compensation, 39 Self-Organizing Map (SOM), 20, 24, 25
predicting business failure, 36 Senhance, 62
quantile regression, 33, 35–36, 37 SERVQUAL model, 81–82, 131
random forest regression, 33 shareholders’ interest and machine
in revenue model, 35 learning, 21
ridge regression, 32, 34, 35, 36 shipping applications of machine learning,
in risk analysis, 35 13–14
simple linear regression, 31, 34 simple linear regression model, 31, 34
supply chain fexibility, 39–40 single port access (SPA) surgical robot,
support vector regression, 33, 39 61–65, 62
types of models, 31–34 skin modifcation robots, 59
and value at risk, 35–36 soccer league competition algorithm, 22
rehabilitation robots, 59 social-identity customer experience,
reinforcement learning algorithm, 16 122–123
research data processing, 7 social welfare and artifcial intelligence,
retail sector and artifcial intelligence, 133 133
revenue model and regression models, 35 soft robotics, 77
ridge regression, 32, 34, 35, 36 statistical analysis, 3
risk analysis and regression models, 35 steel industry using machine learning, 23
ROBODOC, 60 stepwise regression model, 36, 38
robotically assisted surgical (RAS) devices, sugar industry using machine learning,
see surgical robots 23
robotics supervised learning algorithm, 15–16
in medical industry, 57–72, 59, 64, 65 supply chain fexibility and regression
soft, 77 models, 39–40
robotics assisted surgery, 58–72 supply chain management and artifcial
benefts, 60–61 intelligence, 14, 91, 133–134
growth of, 64–66 supported vector machine (SVM), 19, 22
 Index ◾ 173

support vector regression, 33, 39 V

surgical robots, 58–60, 61
classifcation of, 59, 61, 61–65, 62 value at risk and regression models, 35–36
increasing acceptance of, 68–69 Verb Surgical, 63, 65
limitations of, 67–68 voice analytics, 77
use of, 65–67
surveillance economy, 156 W
sustainability and machine learning,
17–27, 24 waste management and machine
learning, 19
water cycle algorithm (WCA), 22, 25
T
Tensor Flow, 3 Y
tourism industry, see hospitality industry
YouTube, 104
U
Z
unstructured data, 4
unsupervised learning algorithm, 16 ZEUS Robotic Surgical System, 60