githubofandrew / matrix-factorization-implementation

Matrix-Factorization-Implementation

Here, we will be doing an implementation of Matrix Factorization, the mathematical technique underlying modern, advanced hybrid recommender systems. This is meant to be a clean implementation and not production-level code. We have benchmarked it up on interaction tables that are up to 1000000 entries large, I am working on more optimizations to make it feasible for larger matrices.

Dependencies

• NumPy

Additional Info:

Although I have used numpy for this implementation, if we want to use this for production-level usage on a real-world dataset, we should use Pandas and then impute the missing entries with 0's.

Intended for educational use, feel free to use my implementation straight out of the notebook.