Apriori Algorithm Implementation with Python

Project Overview

Implementation of the Apriori algorithm for frequent itemset mining and association rule learning from transactional datasets. This project demonstrates market basket analysis and pattern discovery capabilities using a grocery store dataset.

Features

• Frequent Itemset Mining: Discover itemsets that meet minimum support threshold
• Association Rule Generation: Extract meaningful rules with confidence and lift metrics
• Market Basket Analysis: Identify product relationships and purchasing patterns
• Customizable Parameters: Adjust support, confidence, and lift thresholds

Technical Implementation

Data Preprocessing

def load_dataset_from_csv(file_path):
    # Load transactional data from CSV file
    # Returns list of transactions as lists of items

Core Algorithm Components

• Candidate Generation: Create new candidate itemsets iteratively
• Support Counting: Measure itemset frequency in transactions
• Pruning Strategy: Eliminate infrequent itemsets early
• Rule Extraction: Derive meaningful associations with confidence and lift

Algorithm Parameters

min_support = 0.0045      # Minimum support threshold
min_confidence = 0.2      # Minimum confidence for rules
min_lift = 3              # Minimum lift value
min_length = 2            # Minimum itemset length for rules

Key Results

The algorithm discovered 28 strong association rules including:

Top Association Rules Discovered:

1. spaghetti, frozen vegetables → olive oil

   • Support: 0.57%
   • Confidence: 20.57%
   • Lift: 3.12

2. pasta → shrimp

   • Support: 0.51%
   • Confidence: 32.20%
   • Lift: 4.51

3. herb & pepper → ground beef

   • Support: 1.60%
   • Confidence: 32.35%
   • Lift: 3.29

Business Insights

• Complementary Products: Identified natural pairings like spaghetti with frozen vegetables and olive oil
• Meal Planning Patterns: Discovered common ingredient combinations for popular dishes
• Cross-selling Opportunities: Found products frequently purchased together for strategic placement

Technical Skills Demonstrated

• Algorithm Implementation: Custom Apriori algorithm from scratch
• Data Analysis: Market basket analysis on transactional data
• Statistical Metrics: Support, confidence, and lift calculations
• Python Programming: Efficient data structures and set operations

Applications

• Retail market basket analysis
• Product recommendation systems
• Store layout optimization
• Promotional strategy development
• Customer behavior analysis

This implementation successfully demonstrates the practical application of association rule mining for extracting valuable business intelligence from transactional data.