Tensorflow based Deals Recommendation System
Requirements: Need to install these python libraries Python3 Tensorflow Numpy Pandas py2neo
backfilling data from dataset to db if you are using this for first time you need to change password to "cyclops" for code to work. Use this command for that :server change-password this will ask you to type in old password and new password. To get old default password go to this link: neo4j/neo4j#5444
run this command python3 examples/data_fill.py you can provide args to pick to pick different datasets
after backfill to run any example: python3 filename
Algorithms used: k-truncated SVD un-supervised learning
Dataset Used: MovieLens 2.5M dataset
Filtering Steps:
- filtering users who have purchased less than 20 deals.
Assumptions:
- Choices made by a user is independent of their age, gender and location but do depend on class of deals he has been 'interested' in (which is derived from his history).