1

UNIT-II
Analytical methods and models: Descriptive analytics-overview of its tools and techniques, role in business
analytics process and its importance in business decision making; Predictive analytics-nature and type of
modelling, basics of data mining and machine learning environment, role in business analytics process and its
importance in strategic decision making; Prescriptive analytics: basics of its tools and modelling, role in business
analytics process.

What is Descriptive Analytics?

Descriptive analytics focuses on summarizing and interpreting historical data to gain insights
into events, patterns, and trends in a business. It involves the exploration and examination of
data through various statistical techniques, visualizations, and KPI reports to provide a clear
understanding of what has happened. This provides you a foundation for decision-making,
performance evaluation, and operational improvements.

Four Types of Analytics

Descriptive analytics sets the foundation for the other 3 types of analytics—diagnostic,
predictive, and prescriptive analytics. While descriptive analytics focuses on summarizing past
data to understand what has happened, diagnostic analytics delves deeper to determine why
something happened by uncovering root causes and relationships. Predictive analytics takes it
a step further by using historical data patterns and statistical models to forecast future
outcomes. Lastly, prescriptive analytics utilizes insights from the previous types to provide
recommendations on the best course of action, optimizing your decision-making processes.
 2

Descriptive analytics provides the necessary context and understanding for your organization
to then diagnose, predict, and prescribe actions based on data-driven insights, forming a
comprehensive analytical framework.

How It Works
Descriptive analytics works by analyzing and summarizing historical data to provide insights
into past events, patterns, and trends. This is why it’s more similar to reporting vs analytics as
most people think of it. The process typically involves several steps:

Collect Data: First, you collect relevant data from various operational sources, such as
databases, spreadsheets, or data warehouses. This data can include structured data such as
numerical values, categorical variables, as well as unstructured data like text or images.
Clean & Prep Data: You then clean and process this data to ensure accuracy and consistency.
This step involves handling missing values, removing duplicates, and transforming the data
into a suitable format for analysis. This step can happen before or after the data is loaded into
a data repository.
Explore Data: Apply exploratory data analysis techniques to gain a better understanding of
the dataset. This includes examining summary statistics, distributions, and visualizations to
identify patterns, outliers, and relationships within the data.
Summarize Data: Calculate descriptive statistics to summarize the dataset, providing key
metrics such as mean, median, mode, standard deviation, and percentiles. These statistics offer
a concise overview of the dataset's central tendencies and dispersion.
Visualize Data: Create visual representations, such as charts, graphs, and dashboards, to
present the data in a visually appealing and easily understandable format. Visualizations help
identify trends, patterns, and anomalies more intuitively.
Interpret Data: Interpret the summarized data and visualizations to extract meaningful
insights and observations about the past events and trends. This analysis enables stakeholders
to understand the data's implications and make informed decisions based on the findings.

Advantages and Disadvantages

Descriptive analytics offers several advantages such as providing historical insights,
supporting data-driven decision-making, and identifying patterns and trends. However, it also
has limitations, including the potential for oversimplification, reliance on quality data, and its
reactive nature rather than being proactive in addressing future challenges. Let’s dig in a bit
deeper:

Advantages:
• Provides a historical perspective: Descriptive analytics allows businesses to gain
insights into past events, patterns, and trends, providing a valuable historical context
for decision-making.
• Easy to understand and interpret: Summarized and visualized data makes it
accessible to a wide range of users, enabling easy understanding and interpretation of
the information.
 3

• Facilitates data-driven decision-making: By analyzing and summarizing data, it helps

organizations make informed decisions based on evidence rather than relying solely on
intuition or guesswork.
• Identifies patterns and trends: It uncovers patterns and trends within data, allowing
businesses to identify opportunities, risks, and areas for improvement.
• Supports performance evaluation: By providing metrics and key performance
indicators (KPIs), it aids in evaluating business performance, measuring progress, and
setting benchmarks.
Disadvantages:
• Limited in terms of insights: It primarily focuses on summarizing past data and may
not provide in-depth insights into the underlying causes or future predictions.
• Lack of context: It may present data without providing sufficient context or
explanations, requiring additional analysis and interpretation to derive meaningful
insights.
• Reactive rather than proactive: It deals with historical data, which means it offers
insights into past events but may not proactively address future challenges or
anticipate changes.
• Reliance on quality data: Accurate and high-quality data is crucial and if the
underlying data is incomplete, inaccurate, or biased, the insights derived may be flawed
or misleading.
• Potential oversimplification: Summarizing complex data into simplified metrics and
visualizations can oversimplify the intricacies of the underlying information, potentially
missing important nuances.

What tools are used in descriptive analytics?

In descriptive analytics, various tools are used to effectively analyze and summarize
data. BI tools such as Qlik Sense and Power BI provide the full range of capabilities,
including statistical analysis, data visualization, and data integration. Otherwise,
statistical software packages like R, Python (with libraries such as Pandas and NumPy),
and SPSS are commonly employed to perform data calculations, generate summary
statistics, and conduct statistical tests. Data visualization tools like Tableau, Qlik, and
matplotlib help in creating visually appealing charts, graphs, and dashboards to present
data in a clear and understandable format. Database management systems such as SQL
enable efficient data storage, retrieval, and manipulation. Additionally, spreadsheet
software like Microsoft Excel and Google Sheets are often utilized for basic data analysis
and visualization.

What are descriptive analysis techniques?

Descriptive analysis techniques encompass a range of methods used to
summarize and analyze data in order to gain insights into past events, patterns,
and trends. These techniques include statistical measures, data visualization, and
exploratory data analysis. Statistical measures, such as mean, median, mode,
 4

standard deviation, and percentiles, provide numerical summaries of the data's

central tendencies, dispersion, and distribution. Data visualization techniques
utilize charts, graphs, and dashboards to visually represent the data, making it
easier to understand and identify patterns or anomalies. Exploratory data analysis
involves analyzing summary statistics, distributions, and visualizations to uncover
relationships, correlations, and outliers within the dataset. By employing these
descriptive analysis techniques, businesses can effectively summarize, present,
and interpret their data, enabling better decision-making and actionable insights
based on historical information.

Types of Predictive Analytics Models and their Applications

Introduction
As businesses continue to embrace digital transformation and data-driven
decision-making, predictive analytics has emerged as a game-changer. AI-driven predictive
models allow you to harness the power of historical data and statistical algorithms to
make predictions about the future with unparalleled accuracy in mere seconds.

Whether you’re looking to optimize your marketing campaigns, reduce operational costs,
or improve customer experience, predictive analytics can help you achieve your targets. In
this blog, we’ll explore the various types of predictive analytics models, how they work,
their applications, their benefits, and their limitations. By the end of this blog, you’ll have a
comprehensive understanding of how predictive analytics models can elevate your
business and drive better outcomes. Let’s dive in.
What is Predictive Analytics?

Predictive analytics is a subfield of data science that analyses historical data and makes
predictions about future events or outcomes using statistical algorithms, machine
learning, and other techniques. This method assists organizations in identifying patterns,
detecting anomalies, and forecasting future trends.

Predictive analytics is often leveraged by organizations to:

• Predict customer churn and purchasing behaviour

• Detect potential fraud or security threats

• Improve supply chain operations

• Predict equipment failures or maintenance requirements

• Predict disease risks or patient health outcomes

To read more on predictive analytics, click here for our detailed blog.
 5

Learn about how you can predict and prevent churn with
Voxco Intelligence.
Book a Demo
What are Predictive Analytics Models?
Predictive analytics models are statistical or machine learning algorithms that use
historical data to make predictions about future events or behaviors. These models
analyze large amounts of data to identify patterns and trends, which they then use to
forecast future outcomes.

Predictive analytics model development typically entails several steps, including data
preparation, model selection, feature engineering, and model training and testing. The
goal is to create a model that predicts future outcomes accurately based on a set of input
variables or features.

Finance, healthcare, marketing, and retail are just a few of the industries that can benefit
from predictive analytics modeling. Predictive modeling in business analytics can help in
identifying potential risks and opportunities, optimizing operations, and making more
informed decisions based on data. Read our article on the use of predictive analytics in
business to get more insights.
How Do Predictive Analytics Models Work?

Predictive analytics models identify patterns and trends in historical data that can be used
to predict future outcomes. Several steps are usually involved in the process:

• Step 1: Data Collection

Gathering relevant data from multiple sources is the first step in developing a predictive
analytics model.

• Step 2: Data Preparation

Once the data has been collected, it must be cleaned and transformed into an
analysis-ready format.

• Step 3: Feature Selection

In this step, the most relevant variables or features are selected from the dataset to be
used in the model.

• Step 4: Model Selection

There are various types of predictive analytics models available, including regression,
decision trees, and neural networks. Based on the nature of the problem and the
available data, the appropriate model is chosen.

• Step 5: Model Training

 6

The model is trained on historical data in this step to learn patterns and relationships that
can be used to make predictions.

• Step 6: Model Evaluation

The model is evaluated after it has been trained to ensure that it can make accurate
predictions on new, unseen data.

• Step 7: Model Deployment

Finally, the model is used to make predictions based on new data and provide business
insights.

Leverage the power of AI and ML to create predictive

models and identify at risk customers.
Get Started Today
What are the types of predictive analytics models?

Predictive analytics models are classified into several types, including:

1. Regression models: These models use one or more input variables to predict a
continuous numeric value, such as sales revenue or customer lifetime value.

2. Classification models: these are used to categorize data into one of two or more
groups based on input variables. Fraud detection, customer segmentation, and
spam filtering are a few examples.

3. Time series models: These models are used to predict future values based on
historical trends and patterns in time-series data, such as stock prices, weather
patterns, or website traffic.

4. Clustering models: These models are used to categorize data points based on
similarities in their characteristics or behaviors. Clustering models are often used
for customer segmentation and market basket analysis.

5. Neural network models: These models are used to identify complex patterns in
data and are based on the structure and function of the human brain. Image and
speech recognition, natural language processing, and predictive maintenance are all
common applications for neural network models.

6. Decision trees: These models are used to generate a visual representation of

possible outcomes based on various decision paths. They are frequently used to
assess risk, detect fraud, and analyze customer churn.

7. Ensemble models: combine multiple predictive models to improve accuracy while

reducing the risk of overfitting. Random forests, gradient boosting, and stacking
models are some examples.
 7

These are a few predictive analytics models examples. There are numerous other models,
each with its own set of strengths and weaknesses, that can be used to solve various
types of problems.
What are the 3 Levels of Predictive Modeling?

Predictive modeling is divided into three levels, each with a different level of complexity
and accuracy.

• Level 1: Descriptive modeling

This is the most basic level of predictive modeling, involving the use of historical data to
identify patterns and trends. The goal of descriptive modeling is to comprehend what has
occurred in the past and to provide insights into what is likely to occur in the future based
on that data.

• Level 2: Predictive modeling

This level of predictive modeling entails making predictions about future events or
outcomes using statistical algorithms and machine learning techniques. Predictive models
are trained on historical data to identify patterns and relationships between various data
points. Predictive modeling’s goal is to make accurate predictions about future events
based on historical data.

• Level 3: Prescriptive modeling

The most advanced level of predictive modeling, in which predictive models are used to
make recommendations or decisions. Prescriptive models consider a variety of variables
and factors before recommending the best course of action to achieve a specific goal or
outcome. Prescriptive modeling seeks to improve decision-making processes and
maximize business outcomes.
Applications of Predictive Analytics Models

Predictive analytics models have numerous applications in various industries, including:

• Finance: In finance, predictive analytics models are used to forecast stock

prices, credit risk, and other financial metrics. These models are used by banks
and financial institutions to identify potential risks and opportunities.

• Healthcare: uses predictive analytics models to forecast disease outcomes,

identify high-risk patients, and improve patient outcomes. Hospitals and
healthcare providers use these models to improve patient care while lowering
costs.

• Marketing: Marketing uses predictive analytics models to identify customer

segments, predict customer behaviour, and optimise marketing campaigns.
Businesses use these models to improve customer engagement and sales.
 8

• Manufacturing: Predictive analytics models are used in manufacturing to

predict equipment failures, optimise production processes, and reduce
downtime. Manufacturers use these models to increase efficiency and lower
costs.

• Retail: Retailers use predictive analytics models to forecast customer demand,

optimise inventory levels, and improve pricing strategies. These models are
used by retailers to improve customer satisfaction and increase sales.
Experience the power of AI with Voxco Intelligence.

Receive real-time insights that drive impact within your organization.

Book a Demo
The Advantages of Predictive Analytics Models

Businesses can benefit from predictive analytics models in a variety of ways, including:

• Improved Accuracy: Because predictive models incorporate more data and

can detect more complex relationships, they can provide more accurate
predictions than traditional statistical methods.

• Improved Decision-making: Predictive analytics models can assist businesses

in making more informed decisions based on data-driven insights rather than
intuition or guesswork.

• Increased Efficiency: By automating complex data analysis tasks, predictive

analytics models can help businesses save time and resources.

• Competitive Advantage: By identifying opportunities and making strategic

decisions ahead of their competitors, businesses that use predictive analytics
models can gain a competitive advantage.
The Limitations of Predictive Analytics Models

Although predictive models offer a wide range of benefits, they also a couple limitations
that must be considered. Here are some of the key limitations of predictive analytics
models.

• Accurate predictions rely on accurate and complete data: Predictive

analytics models rely heavily on historical data. If the data used to build the
model is inaccurate or incomplete, it can lead to unreliable predictions.
Additionally, if there are significant changes in the data that the model has not
accounted for, the model’s accuracy may decrease over time.

• Flawed or inapt algorithms reduce the model’s accuracy: Secondly,

predictive analytics models are only as good as the algorithms used to create
them. If the algorithms are flawed or not suitable for the data being analyzed,
the predictions may not be accurate. It is essential to select the appropriate
 9

algorithm for the problem being addressed and ensure that it is regularly
updated and refined as new data becomes available.

• Predictive models rarely account for outliers: Predictive analytics models

may not be able to account for unexpected events or outliers. As these events
may not have occurred in the inputted historical data, the predictive model will
not be able to predict outliers accurately. For this reason, it is pivitol to review
the predictions regularly and update the model when necessary to ensure its
accuracy.
Takeaways: What is the best model for predictive analytics?

Predictive analytics models are powerful tools that can help businesses gain insights into
future trends and make informed decisions. Businesses can gain a competitive advantage
in their industry by leveraging historical data and advanced algorithms. Companies in the
financial services, healthcare, and retail industries benefit greatly from the use of
predictive analytics software, predictive analytics tools are a type of tool that is useful to
businesses of all sizes.

The AI capabilities of Voxco Intelligence enable you to build predictive models that deliver
predictive insights into customer behavior, make future forecasts, and make data-driven
decisions to improve customer experience.

The best model for predictive analytics depends on the nature of the problem and the
available data. There are various types of predictive models, each with its own set of
advantages and disadvantages, and selecting the right one is dependent on the specific
needs of the business. It is critical to select a model that is appropriate for the problem
and the available data, as well as to continuously evaluate and refine the model in order
to improve its accuracy and performance.
Data Mining:
The origins of Data Mining
 10

✓ Data mining is a discipline with a long history. It starts with the early Data Mining
methods Bayes’ Theorem (1700`s) and Regression analysis (1800`s)
which were mostly identifying patterns in data.
✓ Data mining is the process of analyzing large data sets (Big Data) from different
perspectives and uncovering correlations and patterns to summarize them into useful
information.

✓ Nowadays it is blended with many techniques such as artificial intelligence, statistics,

data science, database theory and machine learning.
✓ Increasing power of technology and complexity of data sets has lead Data Mining to
evolve from static data delivery to more dynamic and proactive information deliveries;
from tapes and disks to advanced algorithms and massive databases.
 11

✓ In the late 80`s Data Mining term began to be known and used within the research
community by statisticians, data analysts, and the management information systems
(MIS) communities.
✓ By the early 1990`s, data mining was recognized as a sub-process or a step within a
larger process called Knowledge Discovery in Databases (KDD) .The most commonly
used definition of KDD is “The nontrivial process of identifying
valid, novel, potentially useful, and ultimately understandable patterns in data”
(Fayyad, 1996).

The sub-processes that form part of the KDD process are;

1. Understanding of the application and identifying the goal of the KDD process

2. Creating a target data set

3. Data cleaning and pre-processing

4. Matching the goals of the KDD process (step 1) to a particular data-mining method.

5. Research analysis and hypothesis selection

6. Data mining: Searching for patterns of interest in a particular form ,

including

classification rules, regression, and clustering

7. Interpreting mined patterns

8. Acting on the discovered analysis

 12

Data Mining Tasks:

Data mining, also known as knowledge discovery in data (KDD), is the

process of uncovering patterns and other valuable information from large data sets. Given the
evolution of data warehousing technology and the growth of big data, adoption
of data mining techniques has rapidly accelerated over the last couple of decades,
assisting companies by transforming their raw data into useful knowledge.

Data mining functionalities are to perceive the various forms of patterns to be identified in
data mining activities. To define the type of patterns to be discovered in
data mining activities, data mining features are used. Data mining has a wide application
for forecasting and characterizing data in big data.

Data mining tasks are majorly categorized into two categories: descriptive and predictive.

1. Descriptive data mining:

Descriptive data mining offers a detailed description of the data, for example- it gives
insight into what's going on inside the data without any prior idea. This demonstrates
the common characteristics in the results. It includes any information to grasp what's
going on in the data without a prior idea.
2. Predictive Data Mining:
This allows users to consider features that are not specifically available. For example,
the projection of the market analysis in the next quarters with the output of the
previous quarters, In general, the predictive analysis forecasts or infers the features of
the data previously available. For an instance: judging by the outcomes of medical
records of a patient who suffers from some real illness.

Key Data Mining Tasks

1) Characterization and Discrimination

• Data Characterization: The characterization of data is a description of the general

characteristics of objects in a target class which creates what are called characteristic
rules.
A database query usually computes the data applicable to a user-specified class and
runs through a description component to retrieve the meaning of the data at various
abstraction levels.
Eg;-Bar maps, curves, and pie charts.
• Data Discrimination: Data discrimination creates a series of rules called discriminate
rules that is simply a distinction between the two classes aligned with the goal class
and the opposite class of the general characteristics of objects.
 13

2) Prediction

To detect the inaccessible data, it uses regression analysis and detects the missing numeric
values in the data. If the class mark is absent, so classification is used to render the prediction.
Due to its relevance in business intelligence, the prediction is common. If the class mark is
absent, so the prediction is performed using classification. There are two
methods of predicting data. Due to its relevance in business intelligence, a prediction is
common. The prediction of the class mark using the previously developed class model and the
prediction of incomplete or incomplete data using prediction analysis are two ways of
predicting data.
14
 15

3) Classification

Classification is used to create data structures of predefined classes, as the

model is used to classify new instances whose classification is not
understood. The instances used to produce the model are known as data from
preparation. A decision tree or set of classification rules is based on such a
form of classification process that can be collected to identify future details,
for example by classifying the possible compensation of the
employee based on the classification of salaries of related employees in
the company.

4) Association Analysis

The link between the data and the rules that bind them is discovered. And two
or more data attributes are associated. It associates qualities that are
transacted together regularly. They work out what are called the rules of
partnerships that are commonly used in the study of stock baskets. To link
the attributes, there are two elements. One is the trust that suggests the
possibility of both associated together, and another helps, which informs of
associations' past occurrence.

5) Outlier Analysis

Data components that cannot be clustered into a given class or cluster are
outliers. They are often referred to as anomalies or surprises and are also very
important to remember.

Although in some contexts, outliers can be called noise and discarded, they
can disclose useful information in other areas, and hence can be very
important and beneficial for their study.

6) Cluster Analysis

Clustering is the arrangement of data in groups. Unlike classification,

however, class labels are undefined in clustering and it is up to the
clustering algorithm to find suitable classes. Clustering is often called
unsupervised classification since provided class labels do not
execute the classification. Many clustering methods are all
based on the concept of maximizing the similarity (intra-class
similarity) between objects of the same class and decreasing the
similarity between objects in different classes (inter-class similarity).
 16

7) Evolution & Deviation Analysis

We may uncover patterns and shifts in actions over time, with such distinct
analysis, we can find features such as time-series results, periodicity,
and similarities in patterns. Many technologies from space science
to retail marketing can be found holistically in data processing
and features.

❖ Machine Learning

✓ Machine learning (ML) is the study of computer

algorithms that can improve automatically through experience
and by the use of data. It is seen as a part of artificial
intelligence. Machine learning algorithms build a model based on
sample data, known as training data, in order to make predictions or
decisions without being explicitly programmed to do so. Machine
learning algorithms are used in a wide variety of applications,
such as in medicine, email filtering, speech recognition, and
computer vision, where it is difficult or unfeasible to
develop conventional algorithms to perform the needed tasks.

✓ A subset of machine learning is closely related to computational

statistics, which focuses on making predictions using
computers; but not all machine learning is statistical learning.
The study of mathematical optimization delivers methods, theory and
application domains to the field of machine learning. Data mining is
a related field of study, focusing on exploratory data analysis through
unsupervised learning. Some implementations of machine learning
use data and neural networks in a way that mimics the working of a
biological brain. In its application across business problems,
machine learning is also referred to as predictive analytics.

❖ How machine learning works

1. A Decision Process: In general, machine learning algorithms are

used to make a
 17

prediction or classification. Based on some input data, which can be

labelled or unlabeled, your algorithm will produce an estimate about a
pattern in the data.

2. An Error Function: An error function serves to evaluate the

prediction of the model. If there are known examples, an error
function can make a comparison to assess the accuracy of the model.

3. A Model Optimization Process: If the model can fit better to the

data points in the training set, then weights are adjusted to
reduce the discrepancy between the known example and the
model estimate. The algorithm will repeat this evaluate and optimize
process, updating weights autonomously until a threshold of accuracy
has been met.

❖ Machine learning methods

Machine learning classifiers fall into three primary categories.

1. Supervised machine learning

Supervised learning, also known as supervised machine learning, is defined

by its use of labelled datasets to train algorithms that to classify data or
predict outcomes accurately. As input data is fed into the model, it
adjusts its weights until the model has been fitted appropriately.
This occurs as part of the cross validation process to ensure that the model
avoids over fitting or under fitting. Supervised learning helps organizations
solve for a variety of real-world problems at scale, such as classifying spam
in a separate folder from your inbox. Some methods used in supervised
learning include neural networks, naïve bayes,
linear regression, logistic regression, random forest, support vector
machine (SVM), and more.

2. Unsupervised machine learning

Unsupervised learning, also known as unsupervised machine learning, uses

machine learning algorithms to analyze and cluster unlabeled
datasets. These algorithms discover hidden patterns or data groupings
without the need for human intervention. Its ability to discover similarities
and differences in information make it the ideal solution for exploratory data
analysis, cross-selling strategies, customer segmentation, image and pattern
recognition. It’s also used to reduce the number of features in a model
 18

through the process of dimensionality reduction; principal component analysis

(PCA) and singular value decomposition (SVD) are two common approaches
for this. Other algorithms used in unsupervised learning include neural
networks, k-means clustering, probabilistic clustering methods, and more.

3. Semi-supervised learning

Semi-supervised learning offers a happy medium between

supervised and unsupervised learning. During training, it uses a smaller
labelled data set to guide classification and feature extraction from a larger,
unlabeled data set. Semi-supervised learning can solve the problem of having
not enough labelled data (or not being able to afford to label enough data) to
train a supervised learning algorithm.

❖ Reinforcement machine learning

Reinforcement machine learning is a behavioural machine learning model

that is similar to supervised learning, but the algorithm isn’t trained using
sample data. This model learns as it goes by using trial and error. A sequence
of successful outcomes will be reinforced to develop the best
recommendation or policy for a given problem.

Machine Learning is a sub-set of artificial intelligence where computer

algorithms are used to autonomously learn from data and information. In
machine learning computers don’t have to be explicitly programmed but can
change and improve their algorithms by themselves.

Machine learning algorithms enable computers to communicate with humans,

autonomously drive cars, write and publish sport match reports, and find
terrorist suspects.

❖ The origins of machine learning

❖ The term machine learning was coined in 1959 by Arthur

Samuel, an
American IBMer and pioneer in the field of computer
gaming and artificial intelligence. Also the
synonym self-teaching computers was used in this time
period. A representative book of the machine learning research during
the 1960s was the Nilsson's book on Learning Machines, dealing
mostly with machine learning for pattern classification. Interest
related to pattern recognition continued into the 1970s, as described by
 19

Duda and Hart in 1973. In 1981 a report was given on using teaching
strategies so that a neural network learns to recognize 40 characters
(26 letters, 10 digits, and 4 special symbols) from a computer terminal.

❖ Tom M. Mitchell provided a widely quoted, more formal definition of

the algorithms studied in the machine learning field: "A computer
program is said to learn from experience E with respect to some class
of tasks T and performance measure P if its performance at tasks
in T, as measured by P, improves with experience E."This
definition of the tasks in which machine learning
is concerned offers a fundamentally operational definition
rather than defining the field in cognitive terms.
This follows Alan Turing's proposal in his paper
"Computing Machinery and Intelligence", in which the
question "Can machines think?" is replaced with the
question "Can machines do what we (as thinking entities) can do?".

❖ Modern day machine learning has two objectives, one is to classify

data based on models which have been developed, and the other
purpose is to make predictions for future outcomes based on
these models. A hypothetical algorithm specific to
classifying data may use computer vision of moles coupled with
supervised learning in order to train it to classify the cancerous moles.
A machine learning algorithm for stock trading may inform the trader
of future potential predictions.
EVOLUTION OF MACHINE LEARNING:
 20

Machine Learning Frameworks

A Machine Learning Framework is an interface, library or tool which allows

developers to build machine learning models easily, without getting
into the depth of the underlying algorithms. Let’s discuss the Top 10
Machine Learning Frameworks in detail:

TensorFlow

Google’s Tensorflow is one of the most popular frameworks today. It is an

open-source software library for numerical computation using data flow
graphs. TensorFlow implements data flow graphs, where batches of data or
tensors can be processed by a series of algorithms described by a graph.

Theano Theano is wonderfully folded over Keras, an abnormal state neural systems
library, that runs nearly in parallel with the Theano library. Keras’ fundamental
favorable position is that it is a moderate Python library for profound discovering
that can keep running over Theano or TensorFlow.

It was created to make actualizing profound learning models as quick and

simple as feasible for innovative work. Discharged under the tolerant MIT
permit, it keeps running on Python 2.7 or 3.5 and can consistently execute on
GPUs and CPUs given the basic structures.

Sci-Kit Learn

Scikit-learn is one of the most well-known ML libraries. It is preferable for

administered and unsupervised learning calculations. Precedents
implement direct and calculated relapses, choice trees, bunching,
k-implies, etc.

This framework involves a lot of calculations for regular AI and data mining
assignments, including bunching, relapse, and order.

Caffe

Caffe is another popular learning structure made with articulation, speed,

and measured quality as the utmost priority. It is created by the Berkeley
Vision and Learning Center (BVLC) and by network donors.
 21

Google’s DeepDream depends on Caffe Framework. This structure is a

BSD-authorized C++ library with Python Interface.

H20

H20 is an open-source machine learning platform. It is an artificial

intelligence tool which is business-oriented and helps in making a decision
based on data and enables the user to draw insights. It is mostly used
for predictive modeling, risk and fraud analysis, insurance
analytics, advertising technology, healthcare, and customer intelligence.

Amazon Machine Learning

Amazon Machine Learning provides visualization tools that help you go

through the process of creating machine learning (ML) models
without having to learn complex ML algorithms and technology.

It is a service that makes it easy for developers of all skill levels to use
machine learning technology. It connects to data stored in Amazon S3,
Redshift, or RDS, and can run binary classification, multiclass categorization,
or regression on the data to build a model.

Torch

This framework provides wide support for machine learning algorithms to

GPUs first. It is easy to use and efficient because of the easy and fast
scripting language, LuaJIT, and an underlying C/CUDA implementation.

The goal of Torch is to have maximum flexibility and speed in building your
scientific algorithms along with an extremely simple process.

Google Cloud ML Engine

Cloud Machine Learning Engine is a managed service that
helps developers and data scientists in building and running superior
machine learning models in production.

It offers training and prediction services that can be used together or

individually. It is used by enterprises to solve problems like
ensuring food safety, clouds in satellite images, responding four
times faster to customer emails, etc.
 22

Azure ML Studio

This Framework allows Microsoft Azure users to create and train models, then
turn them into APIs that can be consumed by other services. Also, you can
connect your own Azure storage to the service for larger models.

To use the Azure ML Studio, you don’t even need an account to try out the
service. You can log in anonymously and use Azure ML Studio for up to eight
hours.

Spark ML Lib

This is Apache Spark’s machine learning library. The goal of this framework
is to make practical machine learning scalable and easy.

What is Prescriptive Analytics?

Prescriptive analytics is the use of advanced processes and tools to analyze
data and content to recommend the optimal course of action or strategy
moving forward. Simply put, it seeks to answer the question, “What should
we do?”

Two factors driving the growth of prescriptive analytics.

Historically, prescriptive analysis required major infrastructure
investments and hard-to-find data science expertise to develop
proprietary algorithms. Today, cloud data warehouses can now
cost-effectively deliver the storage, power, and speed you need. And
modern AutoML tools (automated machine learning) make it easy for you
to build, train, and deploy custom machine learning models.

The 4 Types of Data Analytics

Prescriptive analytics builds upon the three other types of data analytics
which describe the present and make predictions about the future. It then
uses heuristics, machine learning and rule-based systems to make specific
recommendations based on data and probability-weighted projections.
 23

The 4 Types of Data Analytics Question Answered

Descriptive What happened?
Diagnostic Why did it happen?
Predictive What will happen?
Prescriptive What should we do?
Let’s look a bit deeper at the different processes and stages of human input for
each. You can see below that only prescriptive analysis takes you all the way to a
specific recommended next step you should take (“Action”). Certain situations
require human intuition and judgment and in these cases, it can provide decision
support rather than decision automation.

Prescriptive Analytics Benefits

Ultimately, prescriptive analytics helps you make better decisions about what
your next course of action should be. This can involve any aspect of your
business, such as increasing revenue, reducing customer churn, preventing fraud,
and increasing efficiency. Here are the key benefits in more detail:

• Make data-driven, not instinct-driven decisions. Through advanced

algorithms and machine learning, prescriptive analytics recommends a
specific course of action based on a wide variety of factors including
historical and current performance, available resources, and
 24

probability-weighted projections and scenarios. This lowers the chance

for human bias or error.
• Simplify complex decisions. Prescriptive analysis simulates a variety
of scenarios and provides the probability of different outcomes, both
immediate to long term. This makes it much easier for you to not only
understand the specific recommendation from the tool but also know
the probability of a worst-case scenario and incorporate that into your
plans.
• Focus on execution rather than making decisions. Your organization
is likely flooded with data from a wide variety of sources. And the pace
of business today requires you to move fast. The best prescriptive
analytics tools first break down data silos to analyze an integrated data
set and then provide instant, specific recommendations on your best
course of action. This allows you to focus your effort on executing the
plan.
How Prescriptive Analytics Works
The high-level prescriptive analytics workflow is similar to the
traditional machine learning or AI workflow except that instead of leading
to predictive analytics and what-if scenarios, it leads to recommended
actions.

The details of your process will vary depending on your specific use case
and type of data but below is a high-level overview to get you started.

1) Define the question. As with other data analytics or data

science projects, your first step should be to clearly define the problem
you’re trying to solve or which question you’d like to answer. This will
 25

inform your data requirements and allow your prescriptive model to

generate an actionable output.
2) Integrate your data. Next you’ll gather the data you need and prepare
your dataset. To help your model be the most accurate, you should bring in
data representing every factor you can think of. To prepare data for
machine learning (ML) projects such as this you’ll need to do the following:
• Ensure your dataset is correctly labeled and formatted.

• Avoid data leakage and training-serving skew.

• Clean up any incomplete, missing, or inconsistent data.

• Review the dataset after importing to ensure accuracy.

You’ll be working with big data, possibly in real time, so you’ll need to find
the right tools. As stated above, cloud data warehouses can now cost
effectively bring the storage, power, and speed you need.

3) Develop your model. Now you’re ready to build, train, evaluate and
deploy your prescriptive model. You can hire a data scientist to code one
from scratch or you can leverage an AutoML tool to develop a custom ML
model yourself as a citizen data scientist. Either way, this algorithm-based
model will need to ingest a mix of structured data, unstructured data, and
defined business rules. Analytic techniques used in your model may include
simulation, graph analysis, heuristics, optimization, and game theory. You’ll
have to tweak your model in iterations to get it right and you’ll definitely
want to test your model multiple times using new data to see if the
recommendations generated meet what you would expect.
4) Deploy your model. Once you’re confident in its performance, you can
make your prescriptive model available for use. This may be a one-time
project or as part of an on-going production process. For a one-time
project, an asynchronous batch recommendation is probably most
appropriate. If your model will be integral to a larger process in which
other applications depend on fast predictions, a synchronous, real-time
deployment is best. Your model should automatically adjust as new data is
added over time. This will improve the accuracy of the recommendations.
5) Take action. Now you should review the recommendation, decide if it
makes sense to you, and then take appropriate actions. Some situations
require human intuition and judgment and in these cases, prescriptive
analytics should be viewed as decision support rather than decision
automation. Conversely, if your prescriptive model is integrated to a larger
process, the downstream actions may happen automatically. Application
 26

automation triggering action is one of the top 10 BI and data trends this
year.
Prescriptive Analytics Challenges
• Certain situations require human intuition and
judgment. Sometimes you’ll need to bring additional context such
as knowing external market conditions or competitor activities
which may not be included in the algorithm. In these cases,
prescriptive analytics should be viewed as decision support rather
than decision automation.
• Invalid inputs result in invalid outputs. Even the best models will
produce poor recommendations if the data and/or assumptions
built into your model are not accurate.
• Avoid overkill. Not every organization, campaign, or situation
requires the power of prescriptive analysis. In some cases, the
effort of setting it up may not be worth it.
• Models take time to refine. Training and evaluating your model is
a critical step in getting prescription analytics right. Even with
AutoML tools which get you started with a default model, you
should adjust parameters to better fit your specific use case and
test evaluate your model using new data to see if the
recommendations generated meet what you’d expect.