topic Connecting Test data set to Interactive Grouping in EM in SAS Data Science

Connecting Test data set to Interactive Grouping in EM

ggfggrr — Sat, 23 Feb 2019 14:37:22 GMT

Is the following sequence of connecting the Traing/Validation/Test data sets to Interactive Grouping is right way to do while building the scorecard?

SAS Code - Splits the data set into Training and Validation

Risk_features_test - is the Test data set

I am wondering for the reason that, since test dataset is also used for interactive grouping, whether this will cause false high performance on the test data set.

Thanks

Re: Connecting Test data set to Interactive Grouping in EM

WendyCzika — Fri, 01 Mar 2019 14:21:21 GMT

Assuming you have the role for your Test data set set to Test, then yes this should be correct.

The IGN node doesn't group based on the test data, so no worries about overfitting to that.

Re: Connecting Test data set to Interactive Grouping in EM

ggfggrr — Fri, 01 Mar 2019 14:33:04 GMT

Thank you so much Wendy, it is really a motivating help in my learning.

Kind regards,

Mari

Re: Connecting Test data set to Interactive Grouping in EM

busrafenerci1 — Sun, 21 Jul 2019 15:10:36 GMT

Hello,

I am wondering validation data is used for interactive grouping? If it is not, how can I force to use validation set also.

Thanks,