Week 10: Advanced Deep Learning

Introduction

Week 10 focuses on developing a Customer Recommendation System using historical customer data and product interactions. The goal is to provide personalized product category recommendations to enhance customer satisfaction and engagement. By analyzing customer purchase patterns, feedback, and interaction metrics, we can identify products each customer is likely to be interested in.

Objectives & Overview

The main objectives of Week 10 are:

  • Load and explore the customer dataset with all necessary fields.
  • Create a User-Item Matrix to represent customer interactions with product categories.
  • Normalize the User-Item Matrix to make the data suitable for similarity calculations.
  • Compute a User Similarity Matrix to identify similar customers.
  • Generate top product category recommendations for each customer based on similarity scores.
  • Save the final recommendations to a CSV file for further analysis or deployment.

Dataset Overview

# Column Names
0Unique id
1channel_name
2category
3Sub-category
4Customer Remarks
5Order_id
6order_date_time
7Issue_reported at
8issue_responded
9Survey_response_Date
10Customer_City
11Product_category
12Item_price
13connected_handling_time
14Agent_name
15Supervisor
16Manager
17Tenure Bucket
18Agent Shift
19CSAT Score

Sample User-Item Matrix

Unique ID Affiliates Books & General merchandise Electronics Furniture GiftCard Home Home Appliences LifeStyle Mobile
0001f06a-bd9a-4888-9cf9-0a094a15eaf2 050000000
00053812-fb74-4294-a9d1-5167c5a36c7d 000000010
00064265-7bc3-4627-8e36-6e60a4895343 000000100

Top Recommendations for Customer 0001f06a-bd9a-4888-9cf9-0a094a15eaf2

Product Category Recommendation Score
Affiliates0
Electronics0
Furniture0
GiftCard0
Home0
Home Appliences0
LifeStyle0
Mobile0

Summary

  • Total Customers: 17,196
  • Total Product Categories: 9
  • Recommendations saved to: week10_recommendations.csv

Key Metrics

Total Customers: 17,196
Product Categories: 9
CSV Generated: week10_recommendations.csv

Week 10 Milestones

  • Loaded and explored the dataset successfully.
  • Created the User-Item Matrix for all customers and product categories.
  • Normalized data to prepare for similarity computation.
  • Generated User Similarity Matrix and calculated top product recommendations.
  • Displayed top recommendations for each customer.
  • Saved final recommendations in CSV format for future use or deployment.