In FactorizationRecommender is it possible to access the row/column of the derived factor

User 2032 | 9/29/2015, 3:18:49 PM

I was hoping to use the row/column of the factor matrix corresponding to a give user/item id for classification purposes. This would treat the recommender as sort of a dimensionality reduction method.

Comments

User 1207 | 9/30/2015, 4:41:43 PM

Hi JohnnyM,

Yes, it is. You can access the factors of a trained factorization model using m.get("coefficients")