This is the code repository for the paper "Interpretable Matrix Completion: A Discrete Optimization Approach".
The code for the implementation of our algorithm can be found in the file IMC.jl. The function MatrixOptIntFull implements CutPlanes in the paper while MatrixOptInt implement OptComplete.
It has been tested on Julia V1.4, and we provide a demonstration code on how one can reproduce our synthetic experiments.
To replicate the real data experiments, please download the Netflix Prize Data from e.g. https://www.kaggle.com/datasets/netflix-inc/netflix-prize-data, and the corresponding TMDB data using the API endpoint https://www.themoviedb.org/documentation/api?language=en-US.
Then the algorithm can be ran according to the demonstration code in IMC.jl.
If there are any questions, please reach out to Michael Lingzhi Li at [email protected].