0

I have two dataframes A and B. A has an id user_id that is the unique id of the user and each row in this dataframe represents whether or not the user made a purchase. This has a timestamp for each instance of userid appearing in it. each user_id can make multiple purchases of different objects at different times. I have another table that has user_id associated with its history of window shopping along with a timestamp of this instance of window shopping and also contains the specifics of the item that the user_id was window shopping. As you can see there are multiple instances of user_id in this as well

featuretools documentation says

Note that each ft.Relationship must denote a one-to-many relationship rather than a relationship which is one-to-one or many-to-many.

figs_and_nuts
  • 4,870
  • 2
  • 31
  • 56
  • could you make a users entity that both of the purchase entity and the window shopping entity point to? – Max Kanter Aug 27 '19 at 13:02
  • I can. But that would mean one feature vector created for each user_id. But the feature vector for one user_id has to be different based upon timestamp. When he purchased something in april, feature vector generation should consider window shopping till april and when he purchased something in december the feature vector generation should consider window shopping till december – figs_and_nuts Aug 28 '19 at 05:51
  • @MaxKanter - any suggestions? do you want me to explain my situation better? – figs_and_nuts Aug 28 '19 at 10:17
  • yes, any further details would be helpful. also, look at https://docs.featuretools.com/automated_feature_engineering/handling_time.html. using the cutoff_time parameter with a dataframe of user ids and time might be helpful – Max Kanter Aug 29 '19 at 13:11

0 Answers0