Initial Dataset Overview

• Rows: 34,660
• Columns: 9

Data columns (total 9 columns):
 #   Column       Non-Null Count  Dtype
---  ------       --------------  -----
 0   product      27,900 non-null  object
 1   source       34,660 non-null  object
 2   categories   34,660 non-null  object
 3   date         34,621 non-null  object
 4   didPurchase  1 non-null       float64
 5   doRecommend  34,066 non-null  float64
 6   rating       34,627 non-null  float64
 7   reviews      34,658 non-null  object
 8   title        34,654 non-null  object

Dropped Columns

date and didPurchase

• Contain little to no meaningful data.
• Dropped due to irrelevance in current analysis scope.

source

• No missing values, but contains only a single unique value: "Target".
• Does not provide additional descriptive power or segmentation utility.
• Dropped.

Column-Wise Processing

product

• Missing Values: 6,760 records with null product names.
• Reason for drop:
  • Product identification is essential for recommendation and analysis.
  • While inference from title/review/category using LLMs is possible, it's out of scope for this project.
• Action Taken: Dropped records with missing product names.
• Post-cleaning: 60 unique products retained.