lyst / lightfm

A Python implementation of LightFM, a hybrid recommendation algorithm.
Apache License 2.0
4.77k stars 691 forks source link

ValueError: Test interactions matrix and train interactions matrix share 95786 interactions. This will cause incorrect evaluation, check your data split. #553

Open kholoudsaa opened 4 years ago

kholoudsaa commented 4 years ago

Hi

I am trying to create a hybrid recommender using item's features I have a dataset with 478,930 interactions I am quite sure that each user-item pairs available only once in my data, so it should be no interactions shared between train and test sets
However, I keep getting the issue: ValueError: Test interactions matrix and train interactions matrix share 95786 interactions. This will cause incorrect evaluation, check your data split.

How can I solve this? What are the potential reasons for this?

ehsanmeisami commented 2 years ago

I am having the same problem.