Is there a way we can update the model after fitting it once, like
I collect the data
Perform recommender.fit(data) and build the model
Now if I get some new_data or the rating between an item and a user changes
If I perform recommender.fit(new_data), then the old user_factors and item_factors gets overwritten by the new one and the score becomes NaN for many items
Anyway to resolve that or perform incremental model updates?
ankane
commented
2 years ago
Is there a way we can update the model after fitting it once, like
datarecommender.fit(data)and build the modelnew_dataor theratingbetween anitemand auserchangesrecommender.fit(new_data), then the olduser_factorsanditem_factorsgets overwritten by the new one and the score becomesNaNfor many itemsAnyway to resolve that or perform incremental model updates?