Chapter 4: Predictive and Prescriptive Analytics
Fundamentals of Predictive Analytics
Definition and Scope
Predictive Analytics is an advanced branch of data analytics that utilizes historical data,
statistical algorithms, and machine learning techniques to estimate the likelihood of future
outcomes. While Descriptive Analytics explains "What happened?" and Diagnostic Analytics
explores "Why did it happen?", Predictive Analytics answers the question: "What is likely to
happen in the future?"
In business decision-making, predictive analytics serves as a proactive mechanism rather than
a reactive one. By detecting underlying patterns and correlations within historical datasets,
organizations can forecast market trends, customer behaviors, operational demands, and
potential financial risks.
Predictive analytics is a technology that uses statistical data algorithms and relevant machine
learning techniques to predict future outcomes in a given situation based on historical data
records. The main objective of predictive analytics is to determine what is about to take place
based on what has already happened in the past, including other relevant factors.
The benefits of predictive analytics include:
It leads to improved customer retention for businesses to keep bringing in new customers. It
is also necessary to replace existing customers to avoid any significant loss or reduction in
revenue.
Predictive analytics is important for marketers because it helps them identify the right
customers and increases their customer base. It usually results in more profit for their
business, especially in the long term. Such informative and deep insight is possible only
through the technique of predictive analytics which allows businesses to optimise their
marketing and focus on genuine efforts to acquire customers.
Predictive analytics also helps businesses with diverse requirements to segment their
customer base depending on the requirements that are critical to their operations. Using
predictive analytics, businesses can also optimise their existing data and focus on the right
target audience, extend the target segment, and improve opportunities for better markets.
Predictive analytics is also helpful in improving decision-making processes apart from
defining and identifying the most profitable customer segment.
Types of Predictive Modeling
Predictive analytics models are specifically designed to analyze historical data to identify
patterns, and trends, and use this information to forecast future trends. The popular predictive
analytics models are classification, clustering, and time series models −
1. Classification Models
Classification modelling is a type of supervised machine learning model that assigns data
items to predetermined groups or labels. It's used in different applications, including spam
detection and medical diagnosis. The main feature of classification models is to deal with
categorical data. These models categorize data based on historical data and describe
relationships amongst variables in a dataset. For example, to categorize potential customers,
detect fraud, and evaluate credit risk. Different classification models include logistic
regression, decision trees, random forest, neural networks, and Nave Bayes, "spam" vs. "not
spam".
2. Clustering Models
Clustering models are unsupervised machine learning models. They group data based on
similar features. For example, an e-commerce website can use the model to categorize
customers based on common features and marketing strategies. Some of the most commonly
used clustering algorithms are k-means, mean-shift clustering, DBSCAN, EM clustering,
Gaussian Mixture Models (GMM), and hierarchical clustering.
3. Time Series Models
Time series models use various data inputs at a specific time-frequency, such as daily,
weekly, monthly and so on. It is an effective tool for predictive modelling, especially when
working with data with a temporal component. It entails comprehending the underlying
structure of sequentially ordered data points throughout time and applying it to forecast future
values. Time series analysis is commonly used in finance, healthcare, economics, supply
chain management, and meteorology, among others.
4. Regression Models: Used to forecast continuous outcome variables based on one or more
independent variables. Examples include Linear Regression, Multiple Regression,
and Polynomial Regression.
Importance of Predictive analytics:
Predictive analytics enables businesses to be more proactive by identifying trends influencing
strategic decision-making. Forecasts provide a direction to the decision-makers, so
organizations no longer need to rely on assumptions they should think on realistic ground to
make a strategic plan in advance to survive in the competitive world. Some of the key
important aspects of Predictive analytics are as follows –
1. Strategic Decision-making: Predictive analytics enables businesses to
anticipate events in the future, allowing them to act before issues arise or
opportunities are lost. This alters decision-making from reactive to proactive,
resulting in better outcomes.
2. Improves Profit Margins: Predictive analytics can be used to estimate
inventories, find pricing strategies, predict customer traffic, and plan to
maximize sales.
3. Cost Savings and Efficiency: Predictive analytics may help businesses to
optimize operations, reduce expenses and increase efficiency. Companies can
prevent equipment failure in manufacturing by performing proactive
maintenance.
4. Optimizes Marketing Campaigns: Predictive analytics may uncover new
customer insights and forecast behaviours based on inputs, enabling
businesses to customize marketing campaigns, retain important customers, and
capitalize on cross-selling opportunities.
5. Risk Management: Predictive analytics can detect unusual activity, such as
fraudulent transactions, corporate surveillance, or cyberattacks.
6. Optimized Resource Allocation: It enables firms to deploy resources more
effectively by forecasting demand. For example, retailers utilize it to manage
inventory, ensuring that things are available when needed without
overstocking. In human resources, it aids in the prediction of employment
requirements.
How does Predictive Analytics Work?
Data scientists use predictive models to identify correlations between different elements in
selected datasets. Once data collection is complete, a statistical model is formulated, trained,
and modified to generate predictions. The workflow for building predictive analytics
frameworks follows five basic steps −
1. Define the Problem: A prediction starts with defining a problem more
precisely.
2. Collect Data: Data should be gathered from relevant sources like databases,
data warehouses, social media, transactional records etc. overall, Before
predictive analytics models can be applied, data flows have to be identified,
and datasets can be structured in a repository such as a data warehouse like
BigQuery.
3. Pre-process Data: It includes data cleaning and data transformation. Raw
data is often inconsistent, thus it must be cleaned to eliminate noises, missing
values, and outliers. In Data Transformation, data is changed into an analysis-
ready format. This could include normalizing, aggregating, or encoding data to
maintain consistency.
4. Data Exploration: Analysts examine data to detect trends, correlations, and
patterns; it may include data visualization (charts, graphs) to identify
important relationships in the data.
5. Develop Predictive Models: Data scientists use several tools and strategies to
create predictive models, depending on the problem to be solved and the
nature of the data. Some common types of models include Regression Models,
Classification Models, Clustering Models and Time Series Models.
6. Validate Model: Validate the model using Test Data and Cross-Validation
techniques.
7. Model Deployment: Once the model has been validated and performed
successfully, it can be implemented in the real world.
Applications of Predictive Analytics
Predictive analytics can be used to improve operations, increase revenue, and reduce risk in
practically any business or industry, including banking, retail, utilities, the public sector,
healthcare, and manufacturing. Businesses employ predictive analytics to optimize their
operations and determine whether new items are worthwhile investments. Investors use
predictive analytics to determine where to invest their money. Predictive analytics is used by
Internet businesses to fine-tune buy recommendations for their customers and enhance sales.
In this line, augmented analytics employs big data and machine learning when data size
exceeds.
Some of the realistic applications of Predictive analytics are as follows −
1. Banking
Financial services employ machine learning and quantitative methods to forecast prospects
and clients. Banks can use this data to answer issues such as who is most likely to default on
a loan, which clients offer high or low risks, which consumers are most profitable to target
with resources and marketing spend, and what spending is fraudulent.
2. Fraud Detection
Predictive analytics evaluates all actions on a company's network in real-time to detect
anomalies that suggest fraud or other weaknesses.
3. Manufacturing
Forecasting is critical in manufacturing because it allows for the most efficient use of
resources in the supply chain. Critical spokes of the supply chain wheel, such as inventory
management and the work floor, require accurate forecasting to function properly.
4. Marketing and Sales
Marketing and sales teams are well-versed in using business intelligence reports to assess
previous sales performance; however, predictive analytics allows businesses to be more
proactive in how they engage with their clients across the customer lifecycle. Marketing
specialists preparing a new campaign consider how consumers have reacted to the overall
economy. For example, churn projections can help sales teams detect unsatisfied customers
earlier, allowing them to start conversations to increase retention.
5. Stock Traders
When considering whether or not to buy a specific stock or asset, active traders consider a
range of historical measures.
6. Customer Segmentation
Marketers can utilize predictive analytics to adapt content to specific audiences by
segmenting their client base.
7. Healthcare
Predictive analytics in healthcare is used to detect and manage chronically unwell patients, as
well as to track specific infections like COVID-19. Health-employed predictive analytics are
used to examine health records to understand more about how COVID-19 is detected and
managed. The predictive algorithm uses COVID-19 health records or health data to predict
COVID-19 infection patterns. The model produces effective and useful findings like
accurately identify the root causes to the patients with a high risk or survival rate.
8. Risk Reduction
Credit scores, insurance claims, and debt collectors all employ predictive analytics to assess
and predict potential defaults.
9. Operational Improvement
Companies employ predictive analytics models to estimate inventory, manage resources, and
improve efficiency.
10. Human Resources
Human resources employ predictive analytics to improve a different activity, such as
forecasting future workforce skill requirements or identifying variables that contribute to high
staff turnover.
11. Maintenance Forecasting
Organizations use data to forecast when routine equipment maintenance is required and
schedule it before a problem or malfunction occurs.
Advantages of Predictive Modelling
An organization that understands what to expect based on prior trends has a competitive
advantage in managing inventories, labour, marketing campaigns, and most other aspects of
operations.
1. Security
Every company is worried about data security. An integration of automation and predictive
analytics enhances security. Specific patterns associated with suspicious, and typical end-user
activity can set off specific security measures.
2. Risk Reduction
In addition to protecting data, most organisations are attempting to decrease their risk
profiles. For example, a creditor can utilize data analytics to determine whether a customer is
at a higher-than-average risk of default. Other companies may employ predictive analytics to
determine whether their insurance coverage is appropriate.
3. Operational Efficiency
More efficient workflows lead to higher profit margins. Predictive analytics can help improve
decision-making processes. Because the analytics process analyzes all elements of customer
behaviour and gains in-depth insights, it is extremely beneficial in improving existing
decision-making processes.
4. Improved Decision-making
Running a business necessitates making informed decisions. Any expansion or addition to a
product line or any type of growth necessitates weighing the inherent risk against the
potential outcome. Predictive analytics can assist experts to frame strategic decision-making
and ensure their survival in a competitive world.
5. Customer Retention
Predictive analytics also enables organizations with different needs to segment their customer
base based on the criteria that are crucial to their operations. Businesses can also use
predictive analytics to optimise their existing data, focus on the proper target audience,
broaden the target group, and boost market opportunities.
Regression Analysis:
Regression is a statistical method used in finance, investing, and other disciplines that
attempts to determine the strength and character of the relationship between one dependent
variable (usually denoted by Y) and a series of other variables (known as independent
variables). Also called simple regression or ordinary least squares (OLS), linear regression is
the most common form of this technique. Linear regression establishes the linear relationship
between two variables based on a line of best fit. Linear regression is thus graphically
depicted using a straight line with the slope defining how the change in one variable impacts
a change in the other. The y-intercept of a linear regression relationship represents the value
of one variable when the value of the other is zero. Non-linear regression models also exist,
but are far more complex.
Regression analysis is a powerful tool for uncovering the associations between variables
observed in data, but cannot easily indicate causation. It is used in several contexts in
business, finance, and economics. For instance, it is used to help investment managers value
assets and understand the relationships between factors such as commodity prices and the
stocks of businesses dealing in those commodities.
Simple Linear Regression
Theoretical Foundation
Regression Analysis is a statistical tool used to quantify the relationship between a
quantitative response variable (dependent variable) and one or more explanatory variables
(independent variables).
When the analysis involves only one independent variable and the underlying relationship
is assumed to be a straight line, it is called Simple Linear Regression.
The Simple Linear Regression Model
The population regression model is expressed as:
Y = beta_0 + beta_1 X + \epsilon$$
Where:
Y = Dependent (Target / Response) Variable
X = Independent (Predictor / Explanatory) Variable
\beta_0 = Y-intercept (the expected value of Y when X = 0)
\beta_1 = Slope coefficient (the change in Y for a one-unit change in X)
\epsilon = Unobserved random error term (capturing variables omitted from the
model)
The estimated sample regression equation (prediction equation) is written as:
\hat{Y} = b_0 + b_1 X
Where hat{Y} (read as Y-hat) is the predicted value of Y, b_0 is the sample intercept, and
b_1 is the sample slope.
Core Assumptions of Simple Linear Regression (LINE)
1. Linearity: The relationship between X and Y is linear.
2. Independence: The error terms (\epsilon) are independent of each other (no
autocorrelation).
3. Normality: The errors are normally distributed with a mean of zero.
4. Equal Variance (Homoscedasticity): The variance of the error terms ($\sigma^2$)
remains constant across all values of $X$.