MChatzakis / DIS-RecommenderSystem

Movie Recommendations over the MovieLens dataset using Matrix Factorization

🎉 Movie Recommendations using Matrix Factorization for MovieLens 🎉

Manos Chatzakis, Hind El-Bouchrifi and Lluka Stojollari

{emmanouil.chatzakis, hind.elbouchrifi, lluka.stojollari}@epfl.ch

Distributed Information Systems, Recommender Systems Project, EPFL

Context

This repository contains a solution to a Kaggle competition of Distributed Information Systems course of EPFL. The competition is about the creation of a Recommendation Engine for the MovieLens dataset, which is automatically evaluated in the competition using RMSE.

Approach

We describe our approach in our report (report/Report.pdf), as well us our final submission notebook (src/kaggle.ipynb). Briefly, we utilize an optimized Matrix Factorization approach to generate user rating predictions.

Contents

• src/

  • Complete code for Collaborative Filtering, Content-Based Filtering, Matrix Factorization and Kaggle Submission

• report/

  • report.pdf: A 2-page report describing our approach

• data/

  • MovieLens data