Our teams code for testing a variety of models on the MovieLens Dataset. All models should be able to run with the correct dependencies installed and by changing the location of the movielens-100k dataset in each notebook file.
We have designed two item similarity models, two user-similarity models, and one hybrid user-similarity model.
The user-similarity and hybrid user-similarity recommenders are accessible from the 'User Similarity Recommender' notebook.
The movie-cosine similarity and correlation recommenders are accessible from the 'Item-Based Cosine Similarity' and 'Correlation - Finding Similar Movies' notebooks.
We have designed an Item-Item Collaborative Filtering (Weighted-Average) model, a Gradient Boosting (Ensemble Learning) model and a Latent Factor (Matrix Factorisation) model.
The Item-Item Collaborative Filtering model is aceesible from the 'Item-Item Prediction' notebook.
The Gradient Boosting model is accessible from the 'XGBoost Rating Classifier' notebook.
The Latent Factor Model (Matrix Factorization) is accessible from the '100kMovLen_CF_Matrix Factorization_ratePredictio' notebook.