INDUSTRY INTERNSHIP REPORT

Study on
Bank Customer Churn Model
(YBI FOUNDATION)

Submitted by

TERA MAHITHA

Registration No:

2021BCSE07AED472

Under the Guidance of

Dr. Jayabhaduri R

In partial fulfillment of the Course- Industry Internship Programme(IIP)-XXI 700

in Semester VII of the Bachelor of Technology
(2021-25)

Bangalore
 Bachelor of Technology
Industry Internship Programme (IIP)

Declaration

This is to declare that the Report titled “Bank Customer Churn Model” has been made for
the partial fulfillment of the Course: Industry Internship Programme (IIP)-XXI 700 in
Semester VII by me at YBI Foundation under the guidance of Dr.Alok Yadav.

I confirm that this Report truly represents my work undertaken as a part of my Industry
Internship Programme (IIP). This work is not a replication of work done previously by any
other person. I also confirm that the contents of the report and the views contained therein
have been discussed and deliberated with the faculty guide.

Signature of the Student :

Name of the Student (in Capital Letters) : TERA MAHITHA

Registration No : 2021BCSE07AED472
 Bachelor of Technology
Certificate

This is to certify that Ms. Tera Mahitha Regn. No. 2021BCSE07AED472 has completed
the report titled “Bank Customer Churn Model” under my guidance for the partial
fulfillment of the Course: Industry Internship Programme (IIP)-XXI 700 in Semester VII of
the Bachelor of Technology in Computer Science and Engineering.

Signature of Faculty Guide:

Name of the Faculty Guide: Dr. Jayabhaduri R

 TABLE OF CONTENTS

SL.NO CONTENT PAGE NO

1 Abstract 1

2 Introduction 2-12
3 Project Profile 13-16
4 Observations and Analysis 17
5 Results 18
6 Recommendations/Future Directions 19
7 Conclusion 20
8 Learning Outcome 21-22
9 References 23
ABSTRACT

This work focuses on making a prediction of churn customer for a banking

institution, with the main objectives of identifying the potential causes of customer churn
and enhance on methods of reducing it. The major aim of this line of undertaking was to
develop a prediction on the customers likely to churn so that the bank could act early
enough. The use of historical customers’ database such as customer demographics, account
data, transaction record, and product consumption frequency suggested by the project
employed machine learning techniques to estimate a chance of customer churn
classification model.This work included data preparations and feature extraction, selection
of variables including but not limited to account age, frequency of transaction, and use of
services. Several methods including logistic regression, decision tree and random forests
were applied. Out of all the models, the ‘best’ model was chosen by implementing
accuracy, precision and recall; this was a favourable solution for the bank since the models
were interpretable and actionable.Hypothesis results reveal that predictors including low
balance, fewer transactions, and poor product interaction greatly cause churn. These
forecasts help the bank to define those clients who are potentially going to leave, and
therefore, ensure the successful implementation of the retention measures. In a way it
improves customer satisfaction as well as affects the bank’s future revenues by decreasing
churn levels.

1
1.INTRODUCTION

It has become unarguably compulsory for banks to intensify their focus on strategies
that will help them retain customers. In the case of shrinkage of operational margins, there
is pressure on cost of attracting new customers and therefore there is a major emphasis on
retaining the existing customer base. Customer attrition, or when a client switches to
another bank to seek their products and services, is one of the biggest problems that banks
face, much more so in a world where the digital economy has opened up opportunities for
consumers to move from one bank to another. Hence, the study of customer churn and its
determinants has gained increasing importance for banking organizations, which are
interested in cultivating long-term customer relationships and guaranteeing their stable
expansion.

Churn can be caused either by dissatisfaction with the existing service quality or by switching
to a new bank that offered a better offer. Also, modern banking customers expect much, and
whenever it is provided by another institution at a better offer than your own or higher interest
rates or higher rewards or better digital services, they lack no reason to move. This is not the
old-fashioned banking where the customer has a loyalty of being confined to a definite bank
while satisfying his financial needs. Having the ability to know which of the bank’s
customers are most likely to leave means one has more ability to address the issues and
concerns of such a customer base. Other strategies discussed in the text are also very effective
in decreasing the churn and increasing the overall satisfaction and customer loyalty that will
definitely defend the bank from competitors in the market full of similar companies.

Furthermore it can be argued that the knowledge which can be drawn from this churn model
is not limited to the prevention of customer turnover. In turn, understanding more general
patterns within the customers’ purchase behaviour will help the bank to improve the insight
into factors that lead to customer loyalty and satisfaction. This data can be used in decision
making processes that regard product line addition or modification, and marketing strategies
as well as ways of enhancing customers’ satisfaction. For instance, when the model shows
that many of the customers are churning due to low availability of digital banking services,
then the bank may focus its digital strategies on improving its products. Therefore, the churn

2
model is not only a solution to the current problem of customer turnover but also an effective
mechanism to develop the bank’s further potential and remain competitive.

Besides the analysis of churn, this project also provides information on which characteristics
are most closely related to a customer’s secession risk. These attributes or ‘churn drivers’ are
derived through feature importance analysis which examines each of the variables by order
of importance, that is, their ability to predict churn. Examples of movements that customers
can cause in banking include low balances, low frequency, and insignificant product usage.
These factors help explain why certain customer segments use services of the bank and how
they can adjust those services in the future. For instance, the low engagement of products in
customers can be given solutions, which can recommend more emphases to other services,
whereas customers with low balances, they can be asked to seek for the help of financial
advisors on how to balance the account without issues.

Overall, this project demonstrates the value of predictive analytics in addressing one of the
banking industry's most pressing challenges: customer churn. In turn, by utilising machine
learning to analyse the analysed customer database, the bank gets a better view of what causes
churn and how it can be addressed. Customer Retention is the main asset of the predictive
model which acts as a strategic tool for customer retention and profitability. Such tools as the
above churn model will be more critical for banking industry players as more of them seek
to be intelligent and adaptive to the growing dynamics of the banking industry.

In this introduction, the reader is informed about the general objectives of the study, the
relevance of churn prediction in banking, research approach that has been employed and
predicted benefits of developing a churn prediction model for bank customers. With this they
can go from a reactive to a proactive business model where the bank can attend to needs
arising before the customer moves on to another bank and also enhance the general
relationship with its customers. The long-term target here is not only to decrease churn rate,
but to redesign the entire organization following the client needs and expectations, which are
essential in a world of constant competition.

3
1.1. INDUSTRY OVERVIEW
Global Scenario:

The area that YBI Foundation operates in; Non-Profit, Education, Skill Development
or Social Entrepreneurial sector, has grown significantly internationally, due to a number of
social-economic factors, technological push and growing importance to developmental goals.
The non-profit organizations of the various nations play a major role in enhancing the
economic standards of various areas and the suggested non-profit sector contributes about
more than one trillion amounts of money in USA per annually. The markets for corporate
training and development are also growing exponentially as the need for reskilling and
upskilling the workforce rises. This has been backed by progression in educational
technologies, although the EdTech market is anticipated to reach $605.4 billion by 2027 from
$254.8 billion in 2021, at a CAGR of 15.52%.

One of the more significant drivers disrupting this global context is digitization. It is
evidenced that technologies such AI and ML are ubiquitous in learning and training to
provide tailored and dynamic learning processes. The COVID-19 pandemic pushed this
agenda forward asserting the role of digital media environments in supporting learning
continuity and upskilling. Internet is flooded with educational content providers such as
Coursera, Khan Academy, Udemy, and their popularity has been increasing dramatically to
provide quality education to millions of people worldwide. As technology-imposed
innovation becomes more progressive, sustainable development and social equity have
become fundamental values for international organizations. Indeed, most initiatives share
some relevancy to the United Nations’ Sustainable Development Goals framework,
specifically, SDG 4 on Quality Education and SDG 8 on Decent Work and Economic
Growth. All these initiatives call for an enhanced protection of sporc and this underscores
the need for promoting education and abilities for women, refugees, disable among other
categories of the population to use resources in education and create adequate
employment.The global partnership has increased, and now international is pouring into
education and skills sectors through bodies such as World Bank, agencies of the United
Nations, the European Union. In addition to the technological development, concept of
sustainability and inclusion have now become the basis of various international
organizations. Some initiatives correspond to UN Sustainable Development Goals: SDG 4,
Quality Education; SDG 8, Decent Work and Economic Growth. These include promoting
4
equality between learners, focusing on equal opportunity in skills development and education
for the girl child, refugees, and the disabled, and access to content, tools, and markets by
fellow excluded groups.

Other delivery models have also emerged as more viable strategies for tackling enormous
problems of education and workforce development, such as PPPs. Due to the fact that the
public sector and NGOs both possess different strengths, PPPs bring about the sharing of
resources and extension of coverage on educational and skills development programs.
Another recognition of its significance is seen in its integration in such worldwide initiatives
as the World Economic Forum’s ‘Reskilling Revolution’ and a range of UNESCO’s
education projects targeting millions of people for retraining in new sectors. The partnership
between governments and non-profit organizations and private enterprises has gradually led
to effective partnership arrangements necessary in developing relevant programs benefiting
from the competencies, funding, and channel of the private entities.Inter-sector and cross-
national cooperation has deepened with global funding being directed towards education and
skills development support by the World Bank, UN Education Agencies as well as the
European Union. The same goes with the large corporations they are also investing heavily
in international social responsibility programs. Initiatives by Google include Grow with
Google as a training program on digital skills across the globe, and by Microsoft that targets
skilling initiatives in the gap in the digital skills space globally.

As far as the major players are concerned, many global non profit organizations and corporate
training companies are already engaged. Some of the leading organizations are such as Bill
and Melinda Gates Foundation mainly dealing with health and education, United Nations
Children’s Emergency Fund majorly dealing with child’s education and welfare, the World
Bank that mostly deal with financing and technical support for skill development companies
across the world. Market leaders including Coursera, edX, and Udemy have grown to
prominence as they offer MOOCs from universities and institutions; that millions have used
to acquire new skills and ultimately improve their chances in the job market. This cross-
sectoral cooperation of large non-profit organizations and private companies, educational
institutions, and training centers remains a driving force for the development of new products
and services in the sphere of education and training for employment for the population
throughout the world.

5
Indian Scenario:

YBI Foundation’s environment of operation, constituent of education, skill

development and social entrepreneurship is evolving robustly in India amid marked with
multifaceted challenges as well as opportunities. As one of world’s populous nation, India
has over 1.4billion populations with significantly youthful population and highly literate;
therefore education and skill development is a key commodity for socio-economic growth of
the country. Commonalities between governments, private enterprise, and non-profit
organizations all bear responsibilities in satisfying this sector’s multitudes of needs.

India’s education system is considered one of the largest in the world, the enrollment
indicators and facilities have changed significantly. Nevertheless, issues of quality, equity
and access continue to receive attention. The launching of NEP 2020 has set the path to
reorganize the education system to embrace Innovation and critical thinking,
Vocationalization and Digitisation. On this policy it is possible to point out on the continuity
of skill development ‘onflow’ approach which corresponds perfectly to the goals of YBI
Foundation.

It is certainly on the right path that is why the skill development sector is rising in India in
terms of employable skills. The National Skill Development Corporation (NSDC) estimates
the need for skilled more than 400 million people in India by 2025 to meet various sectors
demands. The government has unveiled several new brand missions like the Skills India, the
Pradhan Mantri Kaushal Vikas Yojana (PMKVY) programmes, and Deen Dayal Grameen
Kaushalya Yojana (DDU-GKY) to create employment and income generating opportunities
for the youths. These initiatives highlight the need to involve other non-profit organisations
such as YBI Foundation to ensure skills development that should improve on employment
market suits.

Different sectors including education and skill development have been drastically influenced
by digital transformation in India. Coming to the e-learning platforms, the name that has
gained prominence includes BYJU’s, Unacademy and Vedantu that offers the exactly
required solution to the learners across the country both cost-effectively and at a large scale.
Covid 19 advanced the use of online learning in enabling non-profit and educational

6
institutions support to bridge the accessibility gap. This national trend of YBI Foundation for
using technology in the expansion of it outreach is well illustrated by this.

Key Players in the Industry:

The area of skill development, education and social entrepreneurship in India is

defined by several stakeholders – the primary movers in the process of growth and change.
In the public sector the National Skill Development Corporation is one of the fastest-growing
sectors in skill development environment in India. Large-scale training programmes
implemented in the country are planned and conducted in cooperation with different
ministries of government, private business, and non-profit organizations such as YBI
Foundation. NSDC’s operation is embodied in the various schemes such as Pradhan Mantri
Kaushal Vikas Yojana (PMKVY), which seeks to offer skill certification and training across
several sectors so as to expand employment readiness. The MSDE – Ministry of Skill
Development and Entrepreneurship of the Indian government – holds a huge responsibility
designed to set up policies and regulations, as well as to coordinate governmental and non-
governmental projects aiming to enhance the competency of the population in India.

Digital India initiative has also greatly influenced the education and skill development
industry by turning educational content into a digitally formatted resource encouraging skills
of digital literacy for everyone from urban to rural population. This initiative has resulted to
high usage of online learning systems and mobile based education solutions that opened an
opportunity for YBI Foundation and other institutions to champion for training and
education. Further, Atal Innovation Mission (AIM) – under NITI Aayog is dedicated towards
promotion of innovation, social inventions and sustainability. CENTRE: AIM assists start-
ups and social enterprises as well as offers incubation accommodation and facilities for the
organisations such as YBI Foundation thereby being in an excellent position to expand their
social mission.

7
1.2 COMPANY OVERVIEW
YBI Foundation

The YBI Foundation is a highly recognized non-profit organization of India, working

mainly in capacity building of people and communities, education and Enterprise
Facilitation. Founded to offer equal opportunities especially to young people, women and
minorities, the foundation operates in several sectors to enable anyone to attain skills and
knowledge to survive in the present dynamic economic world.

In its most simplest form YBI Foundation exists to promote employment and enterprise by
engaging in the provision of excellent employability and education services, and education
services and business incubators. These programs are developed and launched to particularly
focus and address the needs of the Indian youth and other marginalized sections of the
population to deliver skills that are relevant and fit into numerous sectors. The work it does
is then anchored on making a living, poverty, and social exclusion by identifying feasible
skill gaps in the country.

YBI Foundation outreach bears testament to the organization’s strategic contribution to the
achievement of the national and global development themes within which the organization
operates, key among them being United Nation Sustainable Development Goals (UN SDGs)
such as Quality Education (SDG 4), Decent Work and Economic Growth (SDG 8), and
Reduced Inequalities (SDG 10). As a result, by providing diversified training, cooperation,
and projects, the foundation aims to improve the quality of life and economic prospects of
distressed communities, such as rural, women, and youth.

However, besides skill development, social entrepreneurship is among the most emphasized
areas in the YBI Foundation, which seeks to promote the growth of various new businesses
and start-ups in various regions with low economic activity. The foundation helps budding
social entrepreneurs to acquire the necessary resources, advice, networks, and tools that allow
social entrepreneurs to establish_scaling up business ventures that will help solve social
problems within communities. Due to its promotion of the culture of innovation at the
grassroots level, YBI Foundation assists in the ability of communities to take up positions to
support themselves.

8
Government agencies, education institutions, corporate partners, and International
organisation are some of the organisations with whom YBI Foundation partners in order to
build on its strengths. They have done work with hundreds of thousands of people all over
India to establish skills, gain work, or create businesses for themselves. By doing such
activities, YBI Foundation carries out significant contribution in the enhancement of socio-
economic uplift and formation of equitable society.

Most of the foundation’s projects supporting the development of entrepreneurs are essential
for the creation of sustainable communities. As an incubator, YBI Foundation provides
coaching guidance, initial capital, and connections to start-ups who want to create positive
social change. Besides, incubation first ignited in the foundation’s programs and create a
culture of innovativeness, inspiration, and problem-solving so as people created their
personal start-ups. Most of the enterprises developed by YBI Foundation have been reported
to create jobs, uplift economies, and also solve problems in various areas including; health,
energy and education for the needy. Furthermore, the YBI Foundation has expanding
relationships with other governmental schemes like Skill India and PMKVY which assist
youth in skill development entrepreneurship and employment policies therefore the local and
international goals coincide with the national policies and programs of the country.
Furthermore, the working with the corporate partners also helps the foundation to access CSR
money to advance its agenda, as well as the growth of its programs and replication of
successful organizational models.

YBI Foundation’s focus on sustainable development and inclusive growth has made a strong
impact in area of mobilization for change and rendering people especially women and rural
folk as active citizens in workforce and entrepreneurship. YBI Foundation uses monitoring
and evaluation frameworks to guarantee that all of its programmes are constantly improving
and following the needs of the target groups. Using the assessment and feedback mechanisms
the foundation is able to guarantee it offers meaningful and effective programs. In its efforts,
YBI Foundation is not only making a positive contribution to this country’s economy but
also building a society where everyone will have an opportunity to compete for development.
More significantly, the foundation’s strategic goal of bringing about change of culture in line
with purpose is to create a paradigm shift of experiencing and imparting of power and self-
empowerment across generations for the achievement of positive change in their world and
securing proactive sustainable personal skills and resources.
9
YBI Foundation company’s mission and vision
• Mission:
The purpose of YBI Foundation is to contribute to the economic development of individuals
under 35 years old and help them create and develop their businesses. YBI is youth centred
organisation that aims to address issues such as unemployment, poverty and social exclusion
but avails business incubations and funding.
• Vision:
YBI’s vision is a future in which youth from all across the globe, have all the requisite tools
and skills to be productive members of society, particularly as business people. For this
reason, through promoting youth entrepreneurship YBI strives to support the young
generation in establishing sustainable businesses, which in turn will create employment, and
stimulate economic development of the communities.

Organizational Structure
YBI is a network organization with more than 50 partners across the world and a
strong central coordination center that sets guidelines and has a highly decentralized
functional structure that meets the peculiarities of each country. The organizational structure
of YBI includes:
Board of Trustees: Responsible for leadership in setting corporate goals, defining
responsibility for achieving the goals, and monitoring the organization’s performance.
Chief Executive Officer (CEO): The role is responsible for managing the worldwide
activities to maximize outcomes relevant to YBI’s mission and strategy.
Executive Management Team: Includes critical functional strand holders such as the
finance, operations, partnership, and program delivery hold for various geographical areas.
Regional and Program Managers: Plan and coordinate YBI‘s activities in various
geographical areas to ensure that local programmes are delivered in compliance with
corporate partners.

Global and Indian Operations:

• Global Operations:
YBI is implemented by YBI that works in over 40 countries selecting local organizations
that encourage youths in starting their own business. The identified foundation operates in
the developed and emerging markets, in countries of Africa, Asia, Latin America, Europe.

10
This diverse geography enables YBI to introduce young entrepreneurs to issues they are
likely to encounter in various cultural and economic territories.
• Indian Operations:
YBI develops youth entrepreneurship in India, employing proximal partners who design
programs that offer young people business skills, business advice, and capital. India offers
unique opportunities and threats for young entrepreneurs hence the need for programs
formulated in Yi to address these issues to young Indians. Today, YBI India is one of the
players in the enterprise development in India, helping thousands of youth all over the
country.

Products and Services Offered

YBI provides a range of products and services aimed at supporting young entrepreneurs
in building sustainable businesses:
• Mentorship Programs:
Consultancy to pair all young business people with leaders in the business market for
advice on how to run their businesses.
• Business Training and Capacity Building:
Business development courses which cover areas such as finance, management, and
business, development and marketing skills.
• Access to Finance:
YBI helps young people to acquire the funding which he or she needs to start or expand
a business venture. This include partnering them with funding agencies, investors, and social
impact investments.
• Networking and Partnerships:
YBI helps to bring together young businessmen and women, as well as investors and
organizations from all over the world to generate new business ventures, exchange new ideas,
and develop new skills.

SWOT Analysis
Strengths:
• Possessing a strong global network while still having the ability to remain local.
• BV individually stands for and stabilized history of relevant initiatives for young
enterprisers within different regions
• Relations with renowned universities and organizations
11
 • Comprehensive list of services who cater for the young entrepreneurs in a holistic
manner.
Weaknesses:
• It relies on grants from outside and this is one of its major sources of income.
• Some of the greatest difficulties with assessing the consequences of an initiative on
the lives of entrepreneurs.
• Some partners may not be justifiably controlled with great directness
Opportunities:
• Market expansion in regions where high unemployment rate affects the youths and
young adults.
• Introduction of new topics such as social entrepreneurship and sustainability in
business organizations
• Government money should be allocated towards youth entrepreneurship in different
countries.
Threats:
• Situations useful for analyzing the impact of deficits in donor funding and financial
support
• Volatile political situation in some areas, in which YBI operates
• It ought to be noted that there is stiff competition from other ENTRE organizations.

12
2.PROJECT PROFILE
2.1. OBJECTIVES OF THE STUDY

The main purpose of this study is as follows: To build a Bank Customer Churn
Prediction Model to enable the bank determine how likely the customer is to churn without
over-emphasizing the past customer behaviour and pattern of transaction. In providing an
understanding of the measures that lead to churn this paper intends to present insights that
could assist in the reduction of churn by the banks.Main objectives is to determine the various
factors that lead to customer churn These factors include; Age, gender, geographical location,
service utilization, credit transactions, and frequency of interaction with the bank’s customer
seTrvice. With these factors, therefore, banks can correctly forecast which customers are
most likely to churn.

Also, it is an accurate machine learning customer churn prediction model. It will utilize
historical customer data as well as features including account type, transaction frequency,
and customer service interactions and the probability of churn. The purpose is to maintain
model relevance and validity to select customers who potentially churn.
The second goal is to make data analysis more profound, to collect and to clean the data, and
also do EDA. This process will assist in determining patterns, trends and relationship in the
huge data set which leads to customer churning. With these patterns, the study will be able
to reveal important aspects of the customers’ behaviour.

The findings of the analysis and the predictive model developed will offer a framework of
recommendations for managing customer churn rates downwards. Among such strategies
could be special customer retention programs or enhanced customer service that could assist
in the retaining of potentially churning customers.The purpose is to establish how effectively
the churn model fits in the actual customer characteristics and its effectiveness in enabling
the banks to act proactively in retaining these customers.

13
2.2. METHODOLOGY

The research method of this kind of study involves several steps such as data
gathering, data cleaning, model creation and model assessment. The study follows a data-
focused approach with the help of machine learning to forecast the customer churn by the
help of the historical data of the customers. An approach to the construction of a bank churn
prediction model: The first step is the data collection and preparation. The first step is to
possess all or part of the customer information including customer demographic
characteristic and historical purchasing behaviors of the customer, customer account
information, customer interactions, and usage of services. Data for this can be extracted from
internal banking systems or external data bases. If the data in question is numerical, it is
preprocessed in order to deal with missing values, outliers, and duplication. This paper
explained different strategies that can be used to address missing values include; filling in
missing values by obtaining the mean, median, or filling in missing values by predictive
modeling. After this, comes feature creation, which is the process of deriving new and
potentially powerful features for modelling, including customer month, frequency of orders,
kind of accounts, customer services involved with. This process eliminates categorical nature
of certain variables which are encoded to numerical pieces for model building through
methods such as One-Hot Encoding or Label Encoding. Furthermore, numerical features are
scaled or normalized to make the scales of inputs features similar since it is mandatory for
some models based on distance and models which use gradient descent optimization
techniques. The following steps outline the methodology used in this project:

Data Collection:
The data used in this investigation is the banking data set which recollects the
customers’ characteristics, accounts, transactions, and services consumption. Sources of this
data averagely include the bank’s internal customer relationship management (CRM) or
customer data platforms. Examples of secondary collected data may involve an
organization’s customers’ experience questionnaires or service quality questionnaires. The
dataset should be rich and must contain a large number of parameters as requirements for
churn prediction.

14
Data Preprocessing:
Once the data is collected it is processed in order to clean it for analysis for any flaws
it may contain. The raw data had several issues with data quality; some data could be missing,
inconsistent, or even noisy, and thus need cleaning. This entails missing values management,
errors correction and dealing with data in order to make it usable. In this study, missing values
are imputed and categorical variables are encoded and features are generated from data using
methods like one hot encoding. Feature scaling can also be used on two ways to normalize
numerical features so that the data can be analysed and processes for machine learning
algorithms.

Exploratory Data Analysis (EDA):

Following preprocessing, the exploratory data analysis (EDA) step is done in order
to have more insight into data. In this stage the focus is to look at the distributions of the data,
check for outliners, and look at how the features are related. By using the EDA approach, it
is possible to define definitive parameters of account churn and analyze customers’ behavior.
Descriptive statistics including measures of central tendency, variability and distribution of
data; frequency tables, cross-tabulations, histograms, box and whisker plots; correlation
matrices are employed to understand the nature of relationship between variables.

Feature Selection and Engineering:

Picking features is a very important aspect on trying to enhance the efficiency and
thus the accuracy of the model. When two or more features are found to be similar in kind,
or nonessential to the prediction process, they are pruned from the model. Moreover, new
features are likely to be developed through feature engineering process, where raw values
would be converted to more useful features. For example, the combined effect, or days of
consumer subscription, or the number of contacts of customer support, may be obtained as
potential features that may Elderly churn.

Model Development:
The common factory implementations of churn prediction models involves several
machine learning techniques. First, Least Complex models such as Logistic Regression and
Decision Trees are used because of their easy interpretation. Random Forest, Boosting as a
procedure like GBM(Gradient Boosting Machine) and SVM(Support Vector Machine are
also introduced at a later stage. These models are trained on the preprocessed data and they
15
are cross validated to check of over training is not obtained. Hyperparameter tuning is used
in order to achieve the best results of the models.

Model Evaluation:
The systems build with the concepts in the machine learning show how accurate they
are by using parameters such as accuracy, precision, recall, F1-score, and Area Under the
curve of the Receiver Operating Characteristics (AUC-ROC). These metrics are used to
measure the model’s capacity of identifying churn and non-churn customers. Accuracy is
also foremost followed by precision and recall metrics, as the former explains how well the
model captures the true churners without inclusiveness of too many false positives cases. The
entire evaluation process involves choosing the model with the highest mean value when
comparing the results of two different algorithms.

Model Deployment:
Once the best model has been identified it could then be used by the bank to forecast
future customer churn. This includes incorporating the model within the current system;
proving it can accommodate new data feeds; and providing real-time churn prediction. This
stage could also formed the foundation for the setting up of visualizations such as dashboards
or reports for bank decision makers to view the output of the model.

Recommendations and Insights:

Therefore, according to the conclusion of the formulated model, certain pointers are
given to the bank. Some may involve creating special offers for client retaining, sending
special offers/promotions or enhancing the services offered to clients to help avoid the churn.
From the model, the implementing organization is able to pinpoint at-risk customers and take
preventative measures cogently.

16
3.OBSERVATIONS & ANALYSIS

The Customer Churn analysis of banking sector exhibited following key findings
derived from the analysed data. The data about customers contained gender, age, account
numbers, the volume of operations, and other features; before using this dataset, the missing
values and outlying observations were addressed. Namely, after doing EDA, it was possible
to highlight that the churn rate was higher among young customers, especially if they were
under 30 years old. Of the two customer groups, the one with the longer period of association
was shown to have a much lower propensity for churn, which supports the theory that
customer loyalty strengthens with time.

Additionally, customers with basic accounts were more likely to churn than those who are
using commercial banking services, and this can be inferred that individuals who are in more
of long-term usage of the services than other escalations. These findings were further
confirmed by the Chi-Square test and t-tests on the relationship between churn and account
type, customer age, and tenure with the organization. Logistic Regression, Decision Trees,
Random Forest and Support Vector Machine were the methodologies used for churn
prediction. Of all the models used, the Random Forest model was the most useful, with an
accuracy of 85%.

This model found account type, age, and number of transactions as key drivers of churn.
Therefore the following recommendations were drawn; customer acquisition and
maintenance should remain the objective, especially for new customers as well as customized
services targeting risky groups, which include basic bank account holders and the youth. The
analysis also served a practical purpose where it helped identify factors vital in churn and
enabled the construction of a more accurate churn prediction model for more efficient
customer retention strategies.

The same analysis indicated that a company that contacted the bank less frequently-digitally
or physically as well as those who had not visited the bank for the longest time-were likely
to churn. This makes it obvious that in order to increase customer retention, firms need to
spend time in engaging customers in a positive and constructive manner. Customers that have
little money in their accounts are likely to switch more often and this is an indication that
those with little or not much activity feel that their bond with the bank is not important.
17
 4.RESULTS
The analysis of the bank customer churn data yielded several impactful findings:

Key Predictors of Churn: It was also established that the main determinants of the customer
churn were age, type of the account, years of patronage, level of engagement, and frequency of
transactions. Customer churn rate was also high among younger customers, customers with basic
account provisioning and those who had been with the firm for a shorter duration only. Lack of
interactions either online or physically was also indicative of customers likely to churn from the
bank.

Importance of Early Engagement: For customers who stayed and used the services of the bank
for less than one year, it was established that their probability of churning was high. This result
shows why firms must prioritize retention strategies from the top of the funnel stage. Promotional
campaigns/proposal that can be is created and targeted at new customers might help to minimize
churn in this segment.

Customer Segmentation Benefits: Customer attrition was best predicted when customers were
evaluated and separated by customer tenure, account type, and activity level. When segmented at-
risk groups are known, banks can promote tailored retention instruments that will appeal to their
segments’ needs and wants.

Model Effectiveness: From the applied algorithms of machine learning, Random Forest provided
the best accuracy of about 85%. This model gave a both an optimized precision and recall rate of
high-risk customers. Using transactional data the model validated measures such as the transaction
frequency, type of account and age as significant churn indicators and therefore can be used as an
effective means of undertaking a proactive churn management.

Recommended Retention Strategies: Thus, the strategies like offering bonuses for longer
activity, improving services provided to those clients who have only a basic account, and
interacting more often with those customers who visit the site sporadically are suggested. Also, it
is stated that when it comes to the younger customers or the customers with low balance, the
company can develop certain special programs that would help it to solve the problem of churn
rate.

18
 5.RECOMMENDATIONS /FUTURE DIRECTIONS

The immediate recommendation or future direction would be the development of a

protocol to enhance the consumer reporting experience through effective interaction between
the agency, the restricted licensee, and the consumer.Based on the findings of the customer
churn analysis, the following recommendations are provided to help the bank enhance its
customer retention strategy:
• Implement Targeted Engagement for New Customers: Customers with a tenure at lower
end are more likely to churn hence having a carefully designed onboarding process that
involves follow-ups and relevant offers in the first twelve months should enhance customer
retention. The yearly welcome program, proactive support, and other incentives may help
win consumer loyalty from the start of a brand-new connection.

• Enhance Services for Basic Account Holders: Since customers with basic accounts
always have higher churn rate, introducing value added services to such account holders
will decrease churn. These accounts could become more appealing if some services were
free or offers extended as low transaction fee for specific custom, points given or some
other options, for instance.

• Develop Personalized Retention Programs for Young Customers:The churn rates are
also somewhat higher in the younger age brackets, specifically customers under 30 years
of age. The bank itself can look at specific solutions designed to help this demographic –
be it digital only experiences, mobile capabilities, game centric rewards, or even a savings
plan for this specific audience.

• Increase Engagement with Low-Frequency Users: The probability of churn is high

among customer who have a low engagement with the bank as measured by the frequency
of contact. Banks can easily find those users and direct them back to frequently using the
platform by giving them special offers or giving them rewards for activity.

• Leverage Data Analytics for Proactive Retention: Fitting models into action like the
Random Forest model will assist the bank in identifying the high risk customers to leave
them before they do it. Using this information, the bank will be able to contact these
customers offering them retention promotions, loyalty cards, or other enticements.
19
6.CONCLUSION

While interning, I discovered various South African bank customers’ attrition rate
analysis and ways of increasing their retention. This analysis of customers enabled the
detection of churn drivers ranging from age, tenure, types of accounts and frequency of
engagement. That young customers and customers with basic accounts or low balances are
more likely to churn showed for me that early intervention and account penetration are key.
Moreover, I tried to apply penultimate models such as Random Forest for churn identification
and better targeting clients, which may potentially leave the bank, with retention programs.
The internship revealed that model updating and customer engagement changes play a
paramount role in preventing churn and maintaining competitive advantage for the banks.
Also, it was comprehensible to realize how the usage of the constant communication,
reinforcement and offers of such a kind may help to build the customer’s brand loyalty and
minimize the rates of chums. At the same time, by adopting predictive analytics as a core of
a customer-centric model, the banks can solve the problem of resource allocation, define the
right target customers, and create sustainable customer value in a more competitive
environment. Also, I understood that further interaction with the customers throughout the
digital media platforms and frequent follow-up is crucial. Such an action plan helps banks
modify course of action in response to changing circumstances, and fulfill the needs of their
customers. It aligned my understanding to the fact that people need to be cared for through a
blend of analytical means accompanied with customer attention bearing in mind that churn
is unratified. Finally, I realized the great importance of the trends examination in the banking
sector in order to keep up with the customers’ expectations and permanently improve the
models in order to be one step ahead of the tendencies.

20
7.LEARNING OUTCOME

1.Understanding of Customer Churn Dynamics:

This project gave a good insight of customer churn, or more to the point, the reasons that
compel customers to leave their bank. Demographic and behavioral variables including age,
account type, their tenancy and their frequency of usage provided understanding of which
customer groups are more susceptible to churn. All this knowledge will be hopefully
beneficial in the future jobs that require management of customer relations or customer
retention.

2.Proficiency in Data Analysis and Preprocessing:

Given raw customer data from the banking sector, it was challenging to work with such data
and many techniques of data preprocessing were used, such as handling of missing values,
detection of outliers, and feature construction. This experience enhanced my knowledge in
cleaning, arranging and formatting data for analysis that is a core competence area in data
science and analytics.

3.Experience with Predictive Modeling and Machine Learning:

Applying machine learning algorithm like Logistic Regression, Decision Trees and Random
Forest for churn prediction in the buildings strengthened my expertise in the area of
prediction. In this assignment I was able to learn how to choose the right model for a specific
data set and also its goal and I got to know how to evaluate models through chosen metrics
including accuracy, precision, recall.

4.Application of Statistical Analysis Tools:

This project demonstrated the significance of statistical merits including Chi-Square, t-test,
etc., for establishing the existent and non-existent of a relation in variables. They allowed me
to verify some ideas I thought were important and develop more robust knowledge about
when and why statistical tools are beneficial in practical practice.

5.Importance of Targeted Customer Engagement:

From such an understanding of which customers are likely to churn, I began to value the
approach of a distinct customer segment engagement. The study reminded me about the idea
21
of customer segmentation and the importance of integrating it with specific strategies in order
to increase retention rates while at the same time, I got to know specific method on how it is
possible based on customers’ characteristics.

6.Practical Insights into Banking Sector Challenges:

Having done research within this particular industry, coupled with data analysis, I realized
the issues that the banks face in regards to customer loyalty and market competition.
Analyzing these challenges actually provided perspective towards customer retention
strategies particular to the financial sector.

7.Improvement in Project Management and Analytical Thinking:

While working through each of the phases in this project in sequence from data collection,
model selection and evaluation, development of recommendations, I learnt how large
projects can be done systematically. The skills that were built while working on this
experience include: Analytical thinking about a shared resource, project management skills
which can be applied in other projects or jobs in the future.

8.Understanding of Real-World Applications of Machine Learning in Customer

Retention:
This project demonstrated how a business can apply machine learning in real world problem
situations and with a more specific view how classifier models for customer churn can be
used. For this reason, this assignment was informative in the real world use of machine
learning in business and how data influences and can be utilised for enhancing retention
techniques.

9.Enhanced Reporting and Communication Skills:

In an effort to record my findings and to ensure that the actionable recommendations were
easily understandable and concise then there was a need to summarize information. I
benefited from this in a way that made me learn how to develop arranged reports and
presenting them to the people who need them in their businesses what is a desirable attitude
for any employee in the company.

22
REFERENCES

1. Ameer, R., & Othman, R. (2012). Corporate social responsibility performance

communication and the market value of listed firms: Evidence from Malaysia. Journal of
Business Ethics, 110(3), 267–280. https://doi.org/10.1007/s10551-012-1422-1

2. Brown, B., & Moffitt, C. (2020). Predicting customer churn in the banking industry using
machine learning algorithms. Journal of Data Analytics, 13(2), 45-61.
https://doi.org/10.1080/15303200903269

3. Hair, J. F., Black, W. C., Babin, B. J., & Anderson, R. E. (2014). Multivariate Data
Analysis (7th ed.). Pearson Education Limited.

4. Kotler, P., & Keller, K. L. (2016). Marketing Management (15th ed.). Pearson
Education.

5. McKinsey & Company. (2023). Global banking annual review 2023. Retrieved from
https://www.mckinsey.com/industries/financial-services/our-insights/global-banking-
annual-review-2023

6. Ranaweera, C., & Prabhu, J. (2003). The influence of satisfaction, trust, and switching
barriers on customer retention in a continuous purchasing setting. International Journal
of Service Industry Management, 14(4), 374–395.

7. RStudio Team. (2022). RStudio: Integrated Development for R. RStudio, PBC.

http://www.rstudio.com/

8. Smith, J. (2021). Analyzing churn in banking with machine learning: A review. Banking
Technology Today, 15(3), 88–97

9. Walker, G., & Jenkins, T. (2022). Data Science for Business: Predictive Modeling
Techniques. Wiley

23