Basket Analysis

Analysis of associations in purchase transactions

What is Basket Analysis?

Market Basket Analysis identifies buying patterns and discovers which products are often purchased together. It's like finding the 'best friends' of your products!

With this analysis, you discover association rules that reveal hidden trends in customer purchasing behavior.

Usage examples:

  • • Create bundled product promotions that are purchased together
  • • Arrange products nearby on the shelves to boost sales
  • • Customize product recommendations for customers
  • • Identify cross-selling and up-selling opportunities

Quick Start

  1. 1. Prepare your data in CSV format with transactions and products
  2. 2. Upload the file to the upload page
  3. 3. Set the parameters (minimum support, confidence)
  4. 4. Please wait for processing (usually 2-5 minutes)
  5. 5. Analyze the discovered association rules

How to organize your data

Organize your data in a CSV spreadsheet with two columns:

Column 1: Transaction ID

Unique identifier for each purchase/order. For example: 001, 002, 003

Column 2: Product

Name of the purchased product. For example: Bread, Milk, Coffee

Example of transaction spreadsheet:

transaction_id product
001 Pão
001 Leite
002 Pão
002 Manteiga

💡 Important: Each row represents ONE product in ONE transaction. If a transaction has 3 products, there will be 3 rows with the same transaction_id.

Analysis settings

Minimum Support (Min Support)

Minimum percentage of transactions that must contain the product set to be considered relevant.

Example:

0.01 = 1% das transações (padrão recomendado)

Minimum Confidence

Probability that product B will be purchased when product A is purchased.

Example:

0.5 = 50% de confiança (padrão recomendado)

Minimum Lift (Min Lift)

Measures how much better the rule is than a random purchase. Values > 1 indicate positive association.

Example:

1.0 = padrão (aceita todas as regras)

Understanding the results

The analysis returns association rules in the format: 'If you buy A, then you will probably buy B'.

Rules Metrics

Support

Frequency with which products appear together in transactions.

Example: Support of 0.3 = 30% of transactions contain these products

Confidence

Probability of buying B given that A was purchased.

Example: Confidence of 0.8 = 80% chance of buying B when buying A

Lift

Indicates how much better the rule is than chance.

Lift > 1: Positive association | Lift = 1: Independent | Lift < 1: Negative association

⚠️ Practical tip: Focus on rules with high confidence (>0.7) and high lift (>2) to identify the strongest and most useful associations for the business.

Need help? Contact us: contato@grabatus.com