I'm working on a logistic regression model in order to model a relationship and am facing a class-imbalance problem (way too many 0's and not enough 1's). In order to resolve this, I'm planning to use the SMOTE algorithm to create a more balanced dataset.

My question is: Once I model the relationship and generate probabilities from each of the coefficients, do I need to do any further transformations since I fixed the class-imbalance?

That is, do I need to "re-adjust" the "adjusted" model since the coefficients are modeled off of simulated values? Or can I just apply the interpretation from the balanced model?


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.