search

lyst / lightfm

A Python implementation of LightFM, a hybrid recommendation algorithm.
Apache License 2.0
4.66k stars 679 forks source link

Recommendation for users the new items in the dataset #688

Open Natsuhadder opened 1 year ago

Natsuhadder commented 1 year ago

Hello , i've trained a lightFM model on my users/vehicles datasets , but the thing is that everytime i have new items/users that are added into the dataset , and each vehicle has a salingstate variable that indicates wheither it's still available on the purchase list or not , so the thing is that i want to , each time , try to recommend to the users only the new vehicles that are added in the dataset and are available in the purchase list based on the preferences they have in the interaction matrix. I've tried to build a new item_feature_matrix for the new vehicles and put it in the predict parameters but it doesn't work correctly , so i was wondering if there was any solution to this problem as i saw a lot of people asking for it but no one has been answered.

johannmartin95 commented 1 year ago

Hello,

I have a similar problem with my new items as well. Is it possible to predict on a new item without training again the model ? And how can we do it ?

Thanks for your response