Step to follow for recommendation on Movie Lens dataset:

a) Pre-run step:

• Clone the repository
• Install packages and dependencies: pip install -r requirements.txt
• Download "20MI" dataset from MovieLens here: https://grouplens.org/datasets/movielens/
• Place the extracted folder into the data/ of the repository
• Run data_subset.ipynb in the src folder. It generates the subset of data on an annual basis starting from 2015 and going backward.

b) For Exploratory Data Analysis: src/matrix-factorization/EDA_personalization.ipynb

c) For ALS and FM model:

• Run src/ALS.ipynb
• Run recommendation_LightFM_WARP.ipynb

d) For Item-Item collaborative filtering:

• Run commmand to create train, test, user mapping dataset for each annual data subsets: ./src/item-item-colab/data_gen.sh
• Run command to create annoy indexes for each annual data subsets: ./src/item-item-colab/create_annoy.sh
• Run python src/item-item-colab/item_colab_metric_plot.py to generate metric plots in output/

NOTE: Final report addressing business use cases, assumptions, design choices, metrics, plots, algorithms here: https://github.com/AmoghM/IEORE4571-RecSys-Movie-Lens/blob/master/Final%20Report%20Movie%20Lens%20RecSys.pdf

Credits

1. Item-based collaborative filtering recommendation algorithms
2. Apache Spark
3. LightFM
4. Annoy Index