Business Data Management Capstone Project

Improving Performance through Customer Segmentation and Shipping Cost Optimization: A Case Study on Walmart

Submitted by: Kulkarni Amit Dilip

Roll Number: 23f1001947
Program: IITM BS Degree, Indian Institute of Technology, Madras
Submission Date: November 27, 2024

---

Project Overview

This project focuses on analyzing Walmart's retail sales operations in the United States, addressing two key challenges:

1. Shipping Cost Optimization – Reducing shipping expenses by optimizing shipping modes based on order characteristics and priority.
2. Customer Segmentation and Retention – Enhancing customer engagement and retention through targeted marketing strategies based on customer segmentation.

Objectives

• Optimize Shipping Costs to improve profit margins while maintaining timely deliveries.
• Segment Customers to provide tailored marketing strategies and enhance customer satisfaction.

---

Dataset

The project uses the Walmart Retail Sales Data sourced from data.world.

• Size: 1,037,247 rows, 23 features
• Key Features:
  • city: The city where the order was placed
  • customer_age: Age of the customer
  • order_priority: Priority level of the order (Critical, High, Medium, Low)
  • ship_mode: Shipping method (Regular Air, Express Air, Delivery Truck)
  • unit_price: Price per unit of the product
  • shipping_cost: Cost of shipping each order

---

Methodology

1. Data Cleaning & Preprocessing

• Removed invalid or missing data.
• Converted object-type columns to numerical or datetime types.
• Scaled numerical features and encoded categorical variables for machine learning models.

2. Exploratory Data Analysis (EDA)

• Correlation Analysis: Identified significant correlations between shipping cost, unit price, and order quantity.
• Visualization: Used bar charts, scatter plots, and line graphs to understand patterns and relationships in the data.

3. Shipping Cost Optimization

• Model Used: Ridge Regression (L2 regularization)
• Key Insights:
  • Higher-priced items incur higher shipping costs due to handling and logistics.
  • Order quantity has a smaller but notable impact on shipping expenses.
• Performance:
  • Mean Absolute Error (MAE): 968.80
  • R² Score: 0.5623

4. Customer Segmentation

• Clustering Algorithm: K-means clustering
• Features Used: Customer age, order quantity, sales, profit, and product category.
• Cluster Characteristics:
  • Cluster 0: Younger customers, high order quantities, moderate sales.
  • Cluster 4: Mid-age customers, high sales, and interest in Technology and Office Supplies.
  • Other clusters exhibited distinct purchase behaviors and profit levels.

---

Key Findings

1. Shipping Cost Optimization

• Critical Orders: Express Air is cost-effective for urgent deliveries.
• High Priority Orders: Regular Air provides significant savings without compromising delivery speed.
• Medium & Low Priority Orders: Delivery Truck is the most cost-effective option.

2. Customer Segmentation

• Younger customers (Cluster 0) favor Office Supplies, while high-value customers (Cluster 4) prefer Technology.
• Older customers (Clusters 1 and 3) purchase smaller quantities but consistently, indicating potential for loyalty programs.

---

Recommendations

1. Bulk Shipping & Negotiated Rates: Implement bulk shipping and negotiate rates for high-value items.
2. Targeted Marketing:
   • Cluster 0: Introduce loyalty programs and small-business discounts.
   • Cluster 4: Offer exclusive product releases and bulk purchase discounts.
3. Tiered Shipping Policy:
   • Critical Orders: Use Express Air for timely delivery.
   • High Priority Orders: Shift to Regular Air for cost savings.
   • Medium & Low Priority Orders: Use Delivery Truck for bulk items to reduce expenses.

---

Conclusion

This project provides a strategic framework to enhance operational efficiency and customer engagement for Walmart. By optimizing shipping costs and implementing targeted marketing strategies, Walmart can improve profitability and customer satisfaction.

---

Files in the Repository

• Proposal.pdf: Project proposal with background, objectives, and expected outcomes.
• Final_Report.pdf: Detailed final report with analysis, findings, and recommendations.
• Presentation.pptx: Project presentation summarizing the key insights and outcomes.

---

Acknowledgments

Special thanks to data.world for providing the dataset and the faculty of IITM BS Degree for their guidance throughout this project.