Instacart Association Rule Mining

A comprehensive market basket analysis project using the Instacart dataset to discover product purchase patterns and association rules through the Apriori algorithm.

📋 Overview

This project performs association rule mining on Instacart's e-commerce transaction data to uncover meaningful product relationships and co-purchase patterns. The analysis helps identify which products are frequently bought together, enabling data-driven recommendations and business insights.

🎯 Objectives

• Analyze customer purchase behavior using market basket analysis
• Extract frequent itemsets using the Apriori algorithm
• Generate association rules with support, confidence, and lift metrics
• Provide actionable business insights for e-commerce optimization

📊 Dataset

The project uses the Instacart Market Basket Analysis dataset, which includes:

• aisles.csv: Aisle information (134 aisles)
• departments.csv: Department information (21 departments)
• products.csv: Product details (49,688 products)
• orders.csv: Order metadata (3.4M+ orders)
• order_products__prior.csv: Prior order products (32M+ records)
• order_products__train.csv: Training order products (1.4M+ records)

Sample Size

The analysis uses a sample of 20,000 orders containing 210,733 product transactions across 21,948 unique products.

🛠️ Technologies & Libraries

• Python 3.x
• pandas: Data manipulation and analysis
• numpy: Numerical computations
• mlxtend: Apriori algorithm and association rule mining
• networkx: Graph-based analysis (optional)
• matplotlib: Data visualization

📦 Installation

1. Clone the repository:

git clone https://github.com/ParsaHaghighatgoo/instacart-association-rule-mining.git
cd instacart-association-rule-mining

2. Create a virtual environment (recommended):

python -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate

3. Install required dependencies:

pip install -r requirements.txt

🚀 Usage

Run the Jupyter notebook to perform the complete analysis:

jupyter notebook main.ipynb

Analysis Pipeline

1. Data Loading: Load and explore the Instacart dataset
2. Data Preprocessing:
   • Filter orders with at least 2 products
   • Merge product names with transaction data
   • Sample 20,000 orders for analysis
3. Basket Creation: Transform transactions into market baskets
4. Apriori Algorithm: Extract frequent itemsets with different support thresholds
5. Association Rules: Generate rules with confidence and lift metrics
6. Visualization: Create charts and graphs for insights
7. Business Interpretation: Translate findings into actionable recommendations

📈 Key Findings

Frequent Itemsets

• min_support = 0.01: 128 frequent itemsets (max size: 2)
• min_support = 0.05: 7 frequent itemsets (max size: 1)

Top Association Rules

The strongest associations (by lift) include:

• Organic Raspberries → Organic Strawberries (lift ≈ 2.95)
• Strong co-purchase patterns among organic fruits
• High correlation between complementary produce items

Metrics Explained

• Support: Frequency of itemset occurrence in transactions
• Confidence: Conditional probability of consequent given antecedent
• Lift: Strength of association (lift > 1 indicates positive correlation)

💼 Business Applications

1. Product Recommendations: Real-time suggestions based on cart contents
2. Cross-Selling & Bundling: Create "Frequently Bought Together" offers
3. Targeted Promotions: Offer discounts on associated products
4. Website Optimization: Place related products closer together
5. Inventory Management: Forecast demand for associated products

📁 Project Structure

instacart-association-rule-mining/
├── main.ipynb                              # Main analysis notebook
├── requirements.txt                        # Python dependencies
├── README.md                               # Project documentation
├── .gitignore                              # Git ignore rules
├── EC-hw3.pdf                              # Assignment description
├── instacart_association_rule_mining_doc.pdf  # Detailed documentation
├── e_commerce_HW3_dataset/                 # Dataset folder
│   ├── aisles.csv
│   ├── departments.csv
│   ├── products.csv
│   ├── orders.csv
│   ├── order_products__prior.csv
│   └── order_products__train.csv
├── Docs/                                   # Analysis documentation
│   ├── Task3_questions.md
│   ├── Task4_Association Rules Analysis.md
│   ├── Task5_BusinessInterpretation.md
│   ├── Task5_Question2.md
│   └── Task5_Question3.md
└── figures/                                # Visualization outputs
    ├── 1.png
    ├── 2.png
    ├── 3.png
    ├── 4.png
    └── top_items.png

📚 Documentation

Detailed analysis and findings are available in the Docs/ folder:

• Task 3: Conceptual questions about support thresholds
• Task 4: Association rules analysis and metrics
• Task 5: Business interpretation and practical applications

📊 Visualizations

The project includes several visualizations in the figures/ directory:

• Top frequent items bar charts
• Association rule network graphs
• Support vs. confidence scatter plots
• Itemset distribution analysis

🔍 Results

The analysis demonstrates that:

• Lower support thresholds (0.01) reveal more meaningful multi-item patterns
• Organic produce shows strong co-purchase associations
• Lift metric is crucial for identifying truly meaningful relationships
• Association rules can drive significant business value in e-commerce

📄 License

This project is part of an academic assignment for educational purposes.

👤 Author

Parsa Haghighatgoo

🙏 Acknowledgments

• Dataset provided by Instacart
• Course: E-Commerce (EC-HW3)
• Institution: CSE Program, Term 9

📞 Contact

For questions or feedback, please open an issue in the repository.

---

Note: This project is for educational purposes and demonstrates the application of association rule mining techniques in e-commerce analytics.