Including Non-negative Matrix Factorization and fixing RatingPredictionEvaluation

[guedes-joaofelipe]

The NMF was added as a rating prediction algorithm, similar to SVD. Also, whenever evaluating a rating-prediction recommender with item-recommendation metric using the evaluate_recommender method, the result would only be printed. A modification was made in order to return such evaluations, followed by an extra example in the 'examples' folder. Finally, comments were included to improve readability.

[arthurfortes]

Hi, @guedes-joaofelipe!

I appreciate your collaboration and your effort to extend this framework. I'm reviewing your commit and I need to ask you some changes:

Remove or put in gitignore

• Folders: /pycache/
• Files: *.pyc
• .ipynb

I'm not using notebook yet for exemplify the use of recommenders. So if you want to keep your notebook, please create a folder in '/examples/', called notebooks and insert your material. Please not save the checkpoints.

One more time, I want to thanks for your help to improve Case Recommender.

[arthurfortes]

Merged!