For introduction to matrix factorization in context of recommender systems see this piece from IEEE.
Probabilistic interpretation of penalty in matrix factorization can be found in Probabilistic Matrix Factorization.
Here you can find the original GLRM paper.
The project requires the following Python packages (I've checked with 3.5)
- jupyter-notebook
- numpy
- pandas
- seaborn
- h2o
Note that first four packages are available in Anaconda by default.
See Makefile for loading and preparing data. For actual interesting stuff see notebooks folder.
