/Amazon-Recommendation-Review-Status-Prediction

Analysis of the Paper White Kindle dataset and Understanding of the Users Rating Distribution. I have tried to Predict Recommend Status based on the subjective review provided by the user

Primary LanguageJupyter Notebook

Amazon-Recommendation Status Prediction

This project focusses on the following areas :

  1. Analysis of the dataset
  2. Understanding of the User's Rating Distribution
  3. Predict Recommend Status based on the subjective review provided by the user

Approach

Clean the Dataset

  1. Clean Column names
  2. Clean Categories
  3. Clean Keys #Analysis of Data

Transforming Date Time

  1. Parse ReviewDate to [Date and Time]
  2. Parse ReviewDateAdded to [Date and Time]
  3. Parse ReviewDateSeen to [Date and Time]

Likert Scale Analysis :

5 Point NPS Breakdown Ratings from 0-3 : Detractors Ratings from 4 : Passive Ratings from 5 : Promoter

Feature Engineering

Apply NLTK - Sentiment Analysis to find Compound Score

Understanding the Fearures Added

Using TF-IDF and Random Forest to predict Recommendation Status