Market Basket Analysis Project

Overview

This project is a comprehensive Market Basket Analysis using Python, aimed at uncovering valuable insights from transactional data. It involves data cleaning, exploratory data analysis (EDA), and the creation of association rules to identify patterns in customer purchasing behavior. Visualizations are provided to enhance data interpretation.

Features

1. Data Cleaning and Preprocessing:

   • Removed duplicates and missing values.
   • Converted date and time information for time-based analysis.
   • Reset dataset indices for consistency.

2. Exploratory Data Analysis (EDA):

   • Sales trends visualized by the hour.
   • Distribution of the number of items purchased by customers.
   • Identification of top-selling products.

3. Market Basket Analysis:

   • Generated frequent item sets using the Apriori algorithm.
   • Created association rules based on metrics such as confidence, support, and lift.

4. Advanced Visualizations:

   • Bar chart for top-selling products.
   • Scatter plot for top association rules, illustrating lift, support, and confidence.
   • Heatmap of association rules for easier interpretation.

Key Insights

• Sales Trends: Peak sales hours were identified, enabling better resource allocation during busy times.
• Customer Behavior: Most customers purchased a limited range of items, with some outliers buying significantly more.
• Top Products: Highlighted products that drive the most sales volume, aiding inventory and marketing decisions.
• Association Rules: Discovered valuable item combinations, facilitating cross-selling and promotional strategies.

Tools and Libraries Used

• Python: Core programming language.
• Pandas: For data manipulation and cleaning.
• Matplotlib and Seaborn: For creating insightful visualizations.
• Mlxtend: For generating frequent item sets and association rules.
• Heatmaps and Scatter Plots: To visualize the relationships and significance of association rules.

Visualizations

1. Top Selling Products:

2.Association Rules Scatter Plot:

3.Lift HeatMap:

Project Workflow

1. Load Dataset: Imported and cleaned the dataset to ensure accuracy and consistency.
2. EDA: Explored sales patterns and customer purchasing behavior.
3. Generate Frequent Item Sets: Applied the Apriori algorithm to find commonly purchased item combinations.
4. Association Rules: Used metrics like lift and confidence to create actionable rules.
5. Visualize Results: Leveraged advanced visualizations to present findings effectively.

Sample Association Rule

• If a customer buys [Product A], they are 85% likely to also buy [Product B] with a lift value of 2.5, indicating a strong relationship.

Future Work

• Incorporate larger datasets for robust analysis.
• Implement real-time recommendations for e-commerce platforms.
• Explore other machine learning techniques to enhance predictive insights.

Contact

Feel free to reach out for collaboration or questions:
Gurleen Singh