Market Basket Analysis in Data Mining
What is Market Basket Analysis?
Market Basket Analysis (MBA) is a data mining technique used by businesses
to understand what products customers buy together.
For example, if many customers buy bread and butter together, the store can
place them near each other or offer combo deals to increase
sales.
Retailers use customer purchase history to find these patterns and improve
marketing strategies.
Why is it Important?
Earlier, shop owners used handwritten records, which were hard to
analyze.
Now, with electronic Point of Sale (POS) systems, all transactions are
stored digitally. This makes it easy to analyze large amounts of data and
find useful patterns.
How Does Market Basket Analysis Work?
MBA is based on Association Rule Mining, which follows this logic:
- IF a customer buys A, THEN they are likely to buy B
Example:
IF a customer buys bread → THEN they may buy butter
This is written as:
{Bread} → {Butter}
Key Terms
1.Antecedent (IF part): The first item
Example: Bread
2.Consequent (THEN part): The related item
Example: Butter
Types of Market Basket Analysis
1. Descriptive Analysis
Uses past data
Finds patterns only (no prediction)
Example: Customers often buy milk and biscuits together
2. Predictive Analysis
Uses models to predict future behavior
Helps in cross-selling
Example: Customers who buy a phone may later buy accessories
3. Differential Analysis
Compares data across:
- Time periods
- Locations
- Stores
Example: Why customers prefer one platform over another
Algorithms Used in MBA
Market Basket Analysis uses algorithms to find relationships between
items:
- Apriori Algorithm (most commonly used)
- AIS Algorithm
- SETM Algorithm
The Apriori Algorithm finds frequently purchased items and builds
combinations step by step.
Important Measures in MBA
1. Support
Shows how often items appear together in all transactions
Support(A,B)=Number of transactions containing 𝐴 and 𝐵 /Total
transactions
Example:
5000 total transactions
500 contain pen
Support (pen) = 500 / 5000 = 10%
2. Confidence
Shows how likely customers buy B when they buy A
Confidence(A→B)=Transactions containing A /Transactions containing A
and B
Example:
1000 transactions contain both pen and notebook
500 contain pen
Confidence = 1000 / 500 = 20%
3. Lift
Shows how strong the relationship is between two items
Lift=Confidence / Support
Example:
Lift = 20 / 10 = 2
If Lift > 1 → Strong relationship
If Lift < 1 → Weak relationship
Real-Life Examples
1.Retail
Online stores recommend “Frequently Bought Together” items
Supermarkets place related items nearby (e.g., shampoo &
conditioner)
2.Telecom
Bundling services like internet + TV packages
3.Banking (IBFS)
Credit card offers and fraud detection
4.Healthcare
Identifying diseases that occur together Studying symptoms and genetic
patterns
Benefits of Market Basket Analysis
1. Increases Sales
Helps businesses suggest related products and improve
cross-selling
2. Understands Customer Behavior
Gives insights into buying habits
3. Improves Store Layout
Helps decide product placement in stores
4. Better Marketing Campaigns
Identifies which products to promote together
5. Recommendation Systems
Used by platforms to suggest:
- Products
- Movies
- Services
Conclusion
Market Basket Analysis is a powerful technique that helps
businesses understand customer behavior and increase sales. By
identifying product relationships, companies can improve
recommendations, marketing strategies, and overall customer
experience.
