I am referring the resources from SAS on the scorecard development process (Link). I see the Logisitc regression does not do well when there are interactions between the characteristics which are inputs to the model. I did not see the methodologies to handle these. I also understand that Decision tree algorithm handles such issues well.
I am thinking of one approach in which I can at first use Decision tree to choose the best variables and then use Interactive grouping before actually feeding them into Scorecard.
Wish to receive an advise whether it makes sense or any cautions need to be taken while doing it.
Thanks
kind regards,
Mari
