Hi there, Is it possible to get two predictions simultaneously in RapidMiner?

ThamaliThamali Member Posts: 1 Newbie
I am doing a credit risk analysis and first I want to get whether that loan will be paid or not. Then if it not paid, I want to predict the percentage of that non-paid amount. I have attached a screenshot of the columns available in my dataset. Please assist me on this issue?

Answers

  • BalazsBaranyBalazsBarany Administrator, Moderator, Employee, RapidMiner Certified Analyst, RapidMiner Certified Expert Posts: 788 Unicorn
    Hi,

    most algorithms are not designed to predict simultaneously a class (classification, yes/no) and a number (regression). But you can easily create a process that does that in two steps, with two different models.

    Use Set Role during building the model to add the label role to the attribute you are predicting and a different role (e. g. other_label) to the attribute you're "hiding" from the current model. Then switch the roles for the second model building step.

    The question is if your data are good enough to predict the non-paid amount. Credit fails are (hopefully) rare in most data, so you might have only a few examples which might be too few for a good model.

    Regards,

    Balázs

Sign In or Register to comment.