Statistical Analysis and Recommendation System 

based on

Amazon Product Reviews

Pavan Manjunath

Hari Prasath Raman

Arvind Ram Anantharam

Overview

  1. Design
     
  2. EDA and Filtering Techniques
     
  3. Matrix Factorization
     
  4. Results
     
  5. Demo

Design Introduction

  • Collaborative Filtering / Social Filtering
     
  • Personal Filtering
     
  • Product Filtering
     
  • Hybrid Filtering
     
  • Matrix Factorization

Social Filtering

Collaborative Filtering based on Jaccard Similarity.

Generate recommendations based on similar users

Personal Filtering

Price Range of the user's buying pattern

Product Filtering

To avoid suggesting bad movies based on review distribution

Hybrid Filtering

To avoid products far away in the network graph from the movies users liked.

Matrix Factorization

 

  • Given that each users have rated some items in the system, we would like to predict how the users would rate the items that they have not yet rated
     
  • Matrix factorization can be used to discover latent features underlying the interactions between two different kinds of entities
     
  • Firstly, we have a set  of U users, and a set  of M movies. Let  R of size |U x M| be the matrix that contains all the ratings. Lets assume K latent features. Our task, then, is to find two matrices matrices  P(a |U x K| matrix) and  Q (a |K x D| matrix) such that their product approximates R:

 

 

Matrix Factorization

Results

Percentage Increase in Success after each level of filtering

Demo

Interactive online tool

http://b671a323.ngrok.io/

deck

By arvind ram

deck

  • 1,012