This repository serves as a framework to build Recommendation Engine in Python. It supports two types of recommender systems based on availability of user item ratings

• Binary Recommender : Leverages users past behaviour in form of binary data (0/1), representing user interaction with item
• Ratings Recommender : Leverages users provided ratings for items they have interacted
  • Utilizes algorithms from Surprise - a Python scikit library for building and analyzing rating based recommender systems,
  • Integrates use of tracking component of MLFlow - A Platform for ML LifeCycle, in order log and track various hyperparameter tuning and model training experiments.

• Books Recommender

  Data : Woongjin Book Club Dataset provided by Kidaptive.

  Project Overview provides a description of the dataset, data preprocessing steps, different types of colloborative and content based recommender systems built, along with evaluation results for comparision.

  Project Readme describes the steps used to leverage various python modules to build project pipeline.

• Articles Recommender

  Data : Articles shared among companies employees provided by CI&T's Deskdrop platform

  Project Readme describes the steps used to leverage various python modules to build project pipeline.

• Media Recommender

  Data : Compass Dataset provided by Kidaptive.

  Project Overview provides a description of the dataset, data preprocessing steps, different recommender algorithms utilized, along with evaluation results comparing various recommenders built with existing recommender at Kidaptive.

  Project Readme describes the steps used to leverage various python modules to build project pipeline.