Machine Learning (4:0:0:0) BCB602
MODULE 1:
Introduction: What Is Business Analytics?, What Is Machine Learning?,
Machine Learning, AI, and Related Terms, Big Data, Data Science, Why
Are There So Many Different Methods?, Terminology and Notation.
Overview of the Machine Learning Process: Core Ideas In ML, The
Steps in a Machine Learning Project, Preliminary Steps, Predictive
Power and Overfitting, Building a Predictive Model with ASDM
Textbook 1: Machine Learning for Business Analytics: Concepts, Techniques, and
Applications with Analytic Solver Data Mining, 4th Edition - Galit Shmueli, Peter C.
Bruce, Kuber R. Deokar, Nitin R. Patel - ISBN: 978-1-119-82983-6
Chapters 1 & 2
Business Analytics
• Art of bringing in large data & use of data
analysis to support business decisions
– Descriptive, Predictive, Prescriptive analytics
– Improves efficiency and competitiveness
Applications?
Business Analytics is used in marketing, finance, HR,
healthcare, education, and manufacturing to
analyze data, predict outcomes, and support better
decision-making.
Industry Business Problem Analytics Used Real-World Example
Real world Understanding
Descriptive
Analyzing monthly sales
Retail sales reports to identify
examples performance
Analytics
top-selling products
of Business E-commerce
Predicting Predictive
customer demand Analytics
Forecasting product
demand during festive
seasons
Analytics Preventing Predictive
Detecting suspicious
Banking transactions using past
financial fraud Analytics
fraud patterns
Optimizing delivery routes
Reducing delivery Prescriptive
Logistics to minimize fuel cost and
cost Analytics
delivery time
Types of Business Analytics
Descriptive Predictive Prescriptive
Descriptive analytics explains the past,
Predictive analytics forecasts the future, and
Prescriptive analytics recommends the best actions.
Descriptive Analytics – “What
happened?”
• Descriptive analytics focuses on analyzing historical data to understand past
performance.
Purpose
• Summarize and interpret past data
• Identify trends and patterns
Techniques used
• Reports and dashboards
• Data aggregation
• Mean, percentage, charts, KPIs
Example
• Monthly sales report
• Student result analysis of previous semester
Output
• Tables, graphs, summaries
Predictive Analytics – “What is likely
to happen?”
• Predictive analytics uses historical data and statistical / machine learning models to
forecast future outcomes.
Purpose
• Predict future trends or probabilities
• Support planning and risk management
Techniques used
• Regression analysis
• Classification models
• Time series forecasting
• Machine learning algorithms
Example
• Predicting future sales
• Forecasting student performance or dropout risk
Output
• Predictions, probabilities, forecasts
Prescriptive Analytics – “What should
we do?”
• Prescriptive analytics goes a step further by recommending actions to achieve the
best possible outcome.
Purpose
• Suggest optimal decisions
• Optimize resources and processes
Techniques used
• Optimization models
• Simulation
• Decision rules
• AI-based recommendation systems
Example
• Optimal inventory ordering level
• Best timetable allocation to reduce faculty workload
Output
• Actionable recommendations
Type of Key Question Sales & Marketing
Outcome
Analytics Answered Example
Identifying
Analysis of last
Descriptive What top-selling
quarter sales by
Analytics happened? products from last
region and product
year’s data
Forecasting next Predicting
Predictive What is likely to month’s sales customer demand
Analytics happen? based on past for upcoming
sales trends festive season
Recommending Deciding which
discount levels products to
Prescriptive What should be
and promotional promote and how
Analytics done?
strategies to much discount to
maximize sales offer
Case Study 1: Retail Chain Sales Performance
• A national retail chain collects data on monthly sales, customer
footfall, and promotional offers across different regions.
Questions:
• Identify examples of descriptive analytics used by the retail chain.
• How can predictive analytics help in forecasting sales for the next
quarter?
• Suggest prescriptive analytics actions to improve sales in
low-performing regions.
• Which data variables are most important for predicting sales?
• What business decisions can be taken based on the analytics
results?
1. Examples of Descriptive Analytics
Descriptive analytics focuses on what has already happened.
The retail chain can use:
• Monthly sales reports by region, store, and product category
• Customer footfall summaries (average daily/monthly visitors)
• Comparison of sales before, during, and after promotions
2. Role of Predictive Analytics in
Forecasting Next-Quarter Sales
•Using historical sales, footfall, and promotion data to
forecast next-quarter sales
•Identifying seasonal trends (festive seasons, end-of-year
sales)
•Predicting demand changes based on past promotion
effectiveness
Prescriptive Analytics Actions for
Low-Performing Regions
•Introduce region-specific promotions or discounts
•Increase marketing campaigns in low footfall areas
•Optimize product assortment based on regional preferences
•Adjust pricing strategies for competitive regions
Most Important Data Variables for
Predicting Sales
•Customer footfall
•Historical sales data
•Promotional offers and discount levels
•Region and store location
•Seasonality and time (month, quarter)
•Product category and pricing
Business Decisions Based on Analytics
Results
• Decide where to invest more marketing budget
• Plan inventory and supply chain efficiently
• Identify underperforming regions for corrective
action
• Schedule promotions during high-impact periods
• Set realistic sales targets for upcoming quarters
Case Study 2: E-Commerce Marketing Campaign
• An e-commerce company runs online advertisements and email
campaigns and tracks clicks, conversions, and customer purchases.
Questions:
• What descriptive analytics reports can be generated from
campaign data?
• How can predictive analytics be used to estimate customer
purchase probability?
• Suggest prescriptive analytics strategies to improve campaign
conversion rate.
• Which metrics indicate campaign success?
• How does analytics support marketing budget optimization?
Case Study 3: Telecom Company Customer Retention
• A telecom provider faces high customer churn and collects
call usage, billing, and complaint data.
Questions:
• Identify descriptive analytics used to analyze customer
churn.
• How can predictive analytics identify customers likely to
leave?
• Suggest prescriptive analytics actions to reduce churn.
• What role does customer segmentation play in this case?
• How can analytics improve long-term customer loyalty?
Case Study 4: FMCG Product Launch
• An FMCG company plans to launch a new product and
analyzes past product launches, pricing, and customer
preferences.
Questions:
• What historical data is used in descriptive analytics?
• How can predictive analytics estimate demand for the new
product?
• Suggest prescriptive analytics decisions related to pricing
and promotion.
• How does analytics reduce product launch risk?
• Which departments benefit most from these insights?
Case Study 5: Digital Marketing for an Educational Institute
• An engineering college promotes its programs through
social media and online ads.
Questions:
• What descriptive analytics can be used to analyze enquiry
data?
• How can predictive analytics forecast student admissions?
• Suggest prescriptive analytics actions to increase
enrolment.
• Which channels should receive higher marketing budget?
• How can analytics improve ROI of marketing campaigns?
[Link]
ubcdlC
Machine Learning
• Subset of Artificial Intelligence
– Learns patterns from historical data
– Improves automatically with experience
• machine learning (or data mining) refers to business analytics
methods that go beyond counts, descriptive techniques,
reporting, and methods based on business rules.
• “what specific advertisement, or recommended product,
should be shown to a given online shopper at this moment?” Or,
we might be interested in clustering customers into different
“personas” that receive different marketing treatment and then
assigning each new prospect to one of these personas.
Machine Learning Workflow
Data Algorithm Model Prediction
Big Data
• Very large and complex datasets
– Cannot be handled by traditional systems
– Defined by 4V’s
– Massive , complex data sets analyzed for insights,
used for predictive analysis across various fileds…
4V's of Big Data
Volume Volume refers to the amount of data.
Velocity refers to the flow rate---the speed
at which it is being generated and changed.
Velocity
Variety refers to the different types of data
being generated (timestamps, location,
Variety
numbers, text, images, etc.).
Veracity refers to the integrity, accuracy,
quality, consistency & trustworthiness of
Veracity data ensuring it is reliable for data analysis
Data Science
• Interdisciplinary field
– Combines statistics, ML, and computing
– Extracts knowledge from data
• Data science is a mix of skills in the areas of
statistics, machine learning, math,
programming, business, and IT.
Why Many Methods?
• Different problem types
– Different data formats
– Trade-offs between accuracy and interpretability
Model Selection Factors
Data Size Data Type
Accuracy Interpretability