0
$\begingroup$

In order to monitor or calculate PSI, I need to have bins of different features. However, in case of tree model, features value in training sample are continue values, I wonder if I require to call

lgb_model.dump_model()

and then using this info using for splitting the tree branch for bin. And besides, what if some features are used in multiple times in different depth of the tree and have a different bin. Let's say we have features:

 1) user.age > 40 -->yes--> 2) user.edu > 3
                        --> 3) user.gender > 1
                             --> 4) user.edu > 4

In the case above, do we need to have user.edu bin like [0-3][3-4][4+]. Does anyone have a package or functions written already?

I have read this PSI article and this Boost article,

and also searched Google, without luck.

$\endgroup$

Your Answer

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

Browse other questions tagged or ask your own question.