CCW331 - Business Analytics Unit-III: Data Mining and

Predictive Analytics

Anna University Semester Exam 15-Mark Answers

Submitted by: Your Name

Roll Number: Your Roll Number

Department of Artificial Intelligence and Data Science

Date: June 3, 2025

Downloaded from EnggTree.com

1 Question 1: Data Mining and Predictive Analysis Modelling

Explain Data Mining and Predictive Analytics Modelling in detail, including their pro-
cesses, key differences, and applications, with a focus on healthcare. Provide an answer
suitable for a 15-mark Anna University semester exam question in an easy and profes-
sional way.

1.1 Introduction

In the era of big data, organizations leverage advanced analytical techniques to extract
valuable insights and drive strategic decision-making. Data Mining and Predictive Ana-
lytics are critical components of Business Analytics, enabling businesses to uncover hidden
patterns and forecast future trends. Data Mining focuses on discovering patterns in large
datasets, while Predictive Analytics uses these patterns to predict future outcomes. This
answer explores their processes, differences, and applications, particularly in healthcare,
in a clear and professional manner.

1.2 What is Data Mining?

Data Mining is the process of extracting meaningful patterns, correlations, and anomalies
from large datasets using statistical, mathematical, and computational techniques. It
transforms raw data into actionable insights, such as identifying customer purchasing
trends in retail or detecting fraudulent transactions in finance. Data Mining is often
handled by mathematicians, engineers, and AI/ML experts.

1.2.1 Process of Data Mining

The Data Mining process consists of six structured phases:

1. Business/Research Understanding Phase: Define the project objectives and require-

ments from a business or research perspective. For example, a retailer might aim
to increase sales by understanding customer preferences.

2. Data Understanding Phase: Collect data and conduct exploratory data analysis to
understand its structure, quality, and initial insights.

3. Data Preparation Phase: Clean raw data (e.g., remove duplicates, handle missing
values) and transform it into a format suitable for modeling tools.

4. Modeling Phase: Select and apply appropriate modeling techniques, such as regres-
sion, clustering, or classification, and optimize model parameters.

5. Evaluation Phase: Assess the models quality and effectiveness to ensure it meets
the objectives set in the first phase.

1
 6. Deployment Phase: Implement the model in production, ranging from generating
reports to integrating complex data mining processes across departments.

1.3 What is Predictive Analytics?

Predictive Analytics uses statistical techniques, machine learning, and historical data to
forecast future outcomes, such as customer behavior or market trends. It builds on pat-
terns identified through Data Mining to make predictions that drive business strategies.
For instance, a telecom company might predict customer churn to offer targeted retention
incentives.

1.3.1 Process of Predictive Analytics

The high-level steps in Predictive Analytics include:

1. Define Business Goal: Articulate the business objective and identify required data.
For example, improving customer retention by segmenting customers based on be-
havior.

2. Data Collection and Preparation: Gather and preprocess data to ensure quality
and compatibility with predictive models.

3. Model Building: Apply machine learning algorithms (e.g., decision trees, neural
networks) to create predictive models.

4. Performance Measurement: Evaluate the models business impact, such as the suc-
cess of a targeted marketing campaign compared to a general one.

5. Deployment and Monitoring: Deploy the model in production and monitor its
performance to ensure ongoing accuracy.

1.4 Key Differences Between Data Mining and Predictive Analytics

1.5 Applications in Healthcare

Both Data Mining and Predictive Analytics are transformative in healthcare, improving
patient outcomes and operational efficiency.

1.5.1 Data Mining in Healthcare

• Fraud Detection: Identifies anomalies in insurance claims or billing data to prevent

fraud.

• Customer Relationship Management: Analyzes patient data to enhance service

delivery.

2
 Aspect Data Mining Predictive Analytics
Objective Discover hidden patterns and Predict future outcomes based
relationships in data. on historical data and patterns.
Process Six phases: Business Focuses on defining business
Understanding, Data goals, modeling, and measuring
Understanding, Preparation, business impact.
Modeling, Evaluation,
Deployment.
Performance Measured by accuracy of pattern Measured by business impact
Measure identification (e.g., classification (e.g., increased sales or reduced
accuracy). churn).
Techniques Statistical methods, clustering, Machine learning, regression,
association rules. time-series analysis.
Future Trends Advances in Deep Neural Expansion into employee
Networks, tools like Googles retention, predictive policing,
Cloud AutoML. etc.
Example Identifying purchase patterns in Predicting which customers are
an online store. likely to churn.

Table 1: Comparison of Data Mining and Predictive Analytics

• Treatment Effectiveness: Evaluates historical treatment data to identify effective

therapies.

• Example: Data Mining can detect suspicious income declarations among millions
of records, ensuring compliance.

1.5.2 Predictive Analytics in Healthcare (Predictive Medicine)

• Personalized Treatment: Predicts optimal treatment plans for patients (e.g., cancer
patients) based on historical data.

• Resource Planning: Forecasts hospital admission peaks to optimize staffing.

• Risk Assessment: Clusters patients into risk groups for diseases like diabetes, en-
abling preventive care.

• Example: Predictive Analytics can adjust hospital staffing based on predicted pe-
diatric admissions or remind patients to refill prescriptions.

1.6 How Data Mining and Predictive Analytics Work Together

Data Mining and Predictive Analytics are complementary. Data Mining uncovers pat-
terns (e.g., customers buying swimsuits also buy sunglasses), which Predictive Analytics
uses to forecast future behavior (e.g., targeting customers likely to buy sunglasses). This
synergy drives actionable insights across industries.

3
1.7 Future Trends

• Data Mining: Evolving with advanced models like Deep Neural Networks and ac-
cessible tools like Googles Cloud AutoML, enabling pattern discovery with fewer
data points.

• Predictive Analytics: Expanding into new areas like employee retention and predic-
tive policing, with organizations collecting more user data for accurate predictions.
The global Predictive Analytics market is projected to grow from $3.49 billion in
2016 to $10.95 billion by 2022.

1.8 Conclusion

Data Mining and Predictive Analytics are essential for unlocking the potential of big
data. Data Mining extracts hidden patterns, while Predictive Analytics forecasts future
outcomes to drive business impact. In healthcare, they enable fraud detection, person-
alized treatments, and efficient resource planning. By understanding their processes and
differences, organizations can harness these techniques to make data-driven decisions and
stay competitive.

4
2 Question 2: Machine Learning for Predictive Analytics

Explain Machine Learning for Predictive Analytics in detail, including its processes, key
algorithms, applications across industries, and the infrastructure needed for effective im-
plementation. Provide an answer suitable for a 15-mark Anna University semester exam
question in an easy and professional way.

2.1 Introduction

Predictive Analytics leverages data to forecast future outcomes, empowering organiza-

tions to make strategic decisions. Machine Learning (ML), a subset of artificial intelli-
gence, is a cornerstone of Predictive Analytics, enabling algorithms to learn from data
and predict trends without explicit programming. This answer explores how ML drives
Predictive Analytics, its processes, key algorithms, applications, and the infrastructure
required for success.

2.2 What is Predictive Analytics?

Predictive Analytics uses statistical techniques, including Machine Learning and Data
Mining, to analyze historical and current data and predict future events, such as cus-
tomer behavior or market trends. It helps businesses anticipate outcomes and optimize
strategies, such as targeting high-value customers.

2.3 What is Machine Learning?

Machine Learning enables computers to learn from data iteratively, identifying patterns
and making predictions without predefined rules. In Predictive Analytics, ML algorithms
power predictive models, adapting to new data to deliver accurate forecasts, such as
predicting stock prices or customer churn.

2.4 How Machine Learning Enables Predictive Analytics

ML is integral to Predictive Analytics through Predictive Modeling, where algorithms

process data to generate forecasts. ML enhances Predictive Analytics by:

• Automating Pattern Detection: Identifies complex patterns in large datasets, such

as fraud indicators.

• Adapting to New Data: Improves predictions as more data becomes available.

• Handling Diverse Data: Processes structured, semi-structured, and unstructured

data, ideal for big data environments.

5
2.4.1 Process of Predictive Analytics with Machine Learning

1. Define Business Objective: Specify the goal, such as reducing churn or forecasting
sales.

2. Data Collection and Preparation: Gather and clean data to ensure quality and
compatibility.

3. Model Building: Train ML algorithms (e.g., neural networks, decision trees) on

historical data.

4. Evaluation: Test model accuracy using a test dataset and refine as needed.

5. Deployment: Implement the model in production for real-time predictions.

6. Monitoring: Continuously assess and retrain the model to maintain accuracy.

2.5 Types of Predictive Models in Machine Learning

1. Classification Models: Predict categorical outcomes (e.g., will a customer buy?

Yes/No).

2. Regression Models: Predict numerical values (e.g., sales revenue or stock prices).

2.6 Key Machine Learning Algorithms for Predictive Analytics

1. Decision Trees: Split data into branches to predict outcomes, used in customer
segmentation.

2. Regression (Linear and Logistic): Predicts continuous (linear) or categorical (logis-

tic) outcomes, such as house prices or churn probability.

3. Neural Networks: Handle complex, nonlinear relationships, ideal for image recog-
nition or fraud detection.

4. Time Series Algorithms: Forecast continuous values over time, such as hospital
admissions.

5. Clustering Algorithms: Group similar data points, such as customer segments.

6. Naïve Bayes: Uses probability for classification, such as spam detection.

7. Support Vector Machines: Classify data by finding optimal boundaries, used in text
analysis.

8. Ensemble Models: Combine multiple algorithms (e.g., Random Forest) for better
accuracy.

6
2.7 Applications of Machine Learning in Predictive Analytics

ML-driven Predictive Analytics transforms industries by converting data into insights:

1. Healthcare:

• Drug Discovery: ML accelerates drug development (e.g., Pfizer using IBMs

Watson).

• Personalized Treatment: Predicts patient responses to therapies based on ge-

netic data.

• Resource Optimization: Forecasts hospital admission peaks for staffing.

2. Finance:

• Fraud Detection: Identifies fraudulent transactions (e.g., PayPal, Citibank

with Feedzai).

• Investment Opportunities: Predicts market trends for trading decisions.

3. Retail:

• Recommendation Systems: Powers personalized suggestions on Amazon and

Netflix.

• Inventory Management: Predicts demand to optimize stock levels.

4. Travel:

• Dynamic Pricing: Ubers Geosurge model adjusts ride prices based on demand.

• Sentiment Analysis: Analyzes user reviews for brand monitoring.

5. Social Media:

• Personalized Content: Facebooks auto-tagging and LinkedIns job recommen-

dations use ML.

2.8 Developing the Right Environment

To leverage ML in Predictive Analytics, organizations must:

1. Ensure Data Quality: Implement data governance to centralize and clean data.

2. Adapt Processes: Integrate ML models into workflows for efficiency.

3. Choose the Right Model: Select algorithms based on the business problem.

4. Use Advanced Tools: Platforms like SAS support data governance and rapid model
deployment.

7
2.9 Comparison Between Machine Learning and Predictive Modeling

Aspect Machine Learning Predictive Modeling

Definition Algorithms that learn from data Statistical approach to predict
without explicit programming. outcomes using data patterns.
Scope Subfield of AI, used in Predictive Core component of Predictive
Analytics and beyond. Analytics, often uses ML
algorithms.
Focus Pattern recognition and Forecasting specific outcomes for
autonomous learning. business decisions.
Algorithms Decision Trees, Neural Networks, Includes ML algorithms plus
SVM, etc. statistical methods.
Applications Image recognition, NLP, Customer churn prediction, sales
predictive maintenance. forecasting, risk assessment.

Table 2: Comparison of Machine Learning and Predictive Modeling

2.10 Conclusion

Machine Learning is the backbone of Predictive Analytics, enabling organizations to har-

ness big data for accurate forecasting. By using algorithms like decision trees, neural
networks, and ensemble models, businesses can predict customer behavior, optimize op-
erations, and mitigate risks across healthcare, finance, retail, travel, and social media. A
robust data governance framework and appropriate infrastructure are essential for success.
As data grows, ML-driven Predictive Analytics will continue to empower organizations
to make data-driven decisions and gain a competitive edge.