Apply Model (Recommender System)

veveveve Member Posts: 63 Contributor II
edited November 2018 in Help

I can't figure out how does this operator  works (for example for the Recommender System extension):
- when I put new clients in logs I have a message "working, : the client is unknown"
- when Iput only clients that are alo in the training test (just different products) I get another worning (test and training sets are overlapping)

The description in the book that is describing this operator "Item Recommendation Apply Model operator also takes a trained model,and a test/query set as input. The Apply Model operator applies the trainedmodel on the query data and returns the list of the first n ranked items for eachuser in the query set, where n is a user defined parameter"

I do not understand
if I should put known clients because the model will try to predict this new items and the apply model will compare this results with the predicted ones ..
or if I should put unknown clients but then I cant't understand how the performences are calculated..

Also, normally we should have train, validation and test..however we have only train and test in rapidminer..how is this done?

Thank you!!

Best regards,
Sign In or Register to comment.