0
$\begingroup$

I have a data set with only two variables, student id and book id. I have train and test sets and I will make prediction about what book student will get next time. Should I attach dummy variables to studentid and bookid? And what ml algorithm can be use for this problem.

Thanks.

$\endgroup$
0
$\begingroup$

If you had more information about the users and the books, you could try to predict if a user with certain attributes might like a book with supervised learning. But since you know nothing about the users except for the ID, you could go with collaborative filtering: "users who borrowed this book, also borrowed this".

Also you could calculate something like the most requested books overall and combine this global ranking over the books with the one you got from collaborative filtering.

$\endgroup$
  • $\begingroup$ I have only info about ID's. I'm thinking about maybe I can get good pred. results with Neural Network. Thank's for your comment $\endgroup$ – sergio0606 Jan 17 at 14:17
  • $\begingroup$ I doubt that, Sergio. At least for supervised approaches like MLP or deep networks. Using only IDs as a feature, you will only end up learning, what you already know: what books have users borrowed in the past. If you go as you plan, your model will not tell you anything interesting. $\endgroup$ – Elmar Macek Jan 17 at 14:44
  • $\begingroup$ Hi Elmar, I have solved the problem. I have used this method. analyticsvidhya.com/blog/2018/04/… github.com/analyticsvidhya/CPT $\endgroup$ – sergio0606 Jan 22 at 7:07

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.