Prediction Error Function for Boosted Trees - Statistica General Discussion - Statistica - Dell Community

Prediction Error Function for Boosted Trees

Prediction Error Function for Boosted Trees

This question is answered

How do you change the prediction error function used to determine where the cut off should be for the number of trees used in Boosted Trees analysis?  On my computer, it had previously used average squared error and now for some reason it is using average multinomial deviance.  I'm not sure what I did to change that and what I should do to change it back?

Verified Answer
  • The average mutlinomial deviance should be used for classification models, while the average squared error is used for regression models. 

All Replies
  • The average mutlinomial deviance should be used for classification models, while the average squared error is used for regression models.