Logistic Results Inconsistent



We were validating Logistic results month on Month and the decile lifts we were getting were consistent for 3 months but from last month the lifts drop. Any particular reason this could happen, we are using same predictors as earlier. We have analysed all predictors MoM to analyze mean and variance and dropped the ones with high difference.

Any list of diagnostic steps one needs to follow to nail the exact reason why it could be happening.


@Arindam- I think your model has become overfit the training data and not performing well on unseen data.
You can check the cross-validation of your model which well help your model to check the performance of the model on any unseen data.

Hope this helps!



We have been sequentially testing the model like building on Jul-Sep and testing on Oct,building on Aug-Oct and testing on Nov so on till Dec but lift drops suddenly from Jan. Any possible reasons