I am working on a classification problem. I have two models:

  1. Logistic regression model

  2. Random Forest model

For the first model, if I choose the only predictors with p-values<0.05 I will reduce the accuracy, recall, etc. The IV (and WOE) predictors selection are not the same as the predictors that comply with the p-value rule.

On the other hand, Random Forest does not give p-values, only importance metrics, which are similar the IV predictors.

So, which methods should use to select the right predictors so I can compare the two models?

Thanks for reading this and I hope you can help me. I will be glad to give more details.

  • $\begingroup$ What is the goal of your analysis? $\endgroup$
    – Dave
    Commented Aug 19, 2021 at 20:40
  • $\begingroup$ I want to predict clients that they are going to upsell next month. $\endgroup$ Commented Aug 22, 2021 at 9:20

1 Answer 1


I recommend finding the best estimators for each model using a validation set and then comparing the final models, each one with different predictors, on the same dataset.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.