There are a lot written about WoE (Weight of Evidence) transformation for the case of Logistic Regression Classifier. It works great.
The question: can one (or does it make sense) to use this WoE approach for, say, Random Forest or Support Vectors Machine?
I'm not sure that the WoE transformation makes much sense at all, at least is looks dated, maybe time to look into better methods. But: The transformation does seem to be coupled to the binary nature of the response, so can be used with binary logistic regression, but also with other methods made for binary response variables, like classification trees and ... (fill out the dots!)
In fact, to me, WoE looks like a variant of target encoding used with binary response (or target) variables, see Target encoding a categorical variable in a highly imbalanced dataset for binary classification.