Side data in regression/classification

User 570 | 10/22/2014, 11:03:22 AM

Hi,

Is it possible to feed input data as "Side data" for regression/classification in same way as for building recommender? When modelling user/item pairs with regression/classification algorithms the user(item) variables are constant and redundant within user(item).

Thanks, Mindis

Comments

User 19 | 10/22/2014, 2:59:36 PM

Hi Mindis,

No, the side_data argument is not available for the regression and classification methods. I recommend joining your data together first. This might look something like the following:

<pre>train.join(users, on='userid').join(items, on='itemid') </pre>

Cheers, Chris