Recommender systems
Exploration of basic recommender systems as an in-class project using the Booking Crossing Dataset. This data is contained in the files BX-Book-Ratings.csv, BX-Books.csv and BX-Users.csv
Data cleaning
Contained within EDA_cleaning.ipynb. Despite the misleading name, EDA is not included. Number of unique books reduced from around 320,000 to 4,533
Model training
Contained within Model_training.ipynb. Basic implementation of Collaborative filtering, Content based filtering and Matrix factorization. Code for matrix factorization is mostly copied from Albert Auyeung's fantastic repo with very minor edits. Output correlation matrices for collaborative/content based filtering are around 300-400MB and too large to upload but should not take too long to run. Timestamps are included in the notebook for reference.
Recommendation Engines
Contained within Recommendation_engines.ipynb. Examples of how the previous models operate and how hybrid systems utilize the output of multiple models to generate recommendations.