How to implement weighted AUC in RM

aldofcesatti

Hi everyone,

I am currently testing 5 different models that are trying to correctly classify a binary classification problem with an imbalanced dataset. The costs of misclassification are slightly different.  

To select the best of the 5 models that I built I want to implement weighted AUC, but I don't know what is the operator that should use to draw an set up the weighted AUCs in RM. Could you please help me?

Thanks in advance.

Aldo

varunm1

Hello @aldofcesatti

Rapidminer has compare ROC operator, that helps get ROC curves for all the models you want to use and you can compare ROC curves to see the better model.

Hope this helps.

MartinLiebig

I think you need to go for ROC to data, weight the points yourself and build a integral.

Best,

Martin