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?