The repository keeps trace of the working procedure of the recommender system challenge on Kaggle (https://www.kaggle.com/competitions/recommender-system-2023-challenge-polimi).
The problem consists in generating potentially relevant book recommendations for users, given a dataset of user-item interactions.
-
Input/ contains the original data (data_train.csv and data_target_users_test.csv), along with saved k-fold splitting and separate train-validation sets.
-
Workspace/ is cloned from the course repository RecSys_Course_AT_PoliMi with some minor modifications of the code. It contains the implementations of various recommender algorithms and utility functions.
-
Notebooks/ contains part of personal notebook files used for the challenge.