I'm performing a regression using a Gradient Boosting Machine. When comparing the cross-validation predictions with the true values, the Mean Percentage Error is around -6%. However, using the model to predict on the training set yields a positive MPE of around 3.5%. How can I expect my model to behave when presented with new data? Will it be biased upwards or downwards? Or rather, am I doing something wrong?

  • $\begingroup$ For (almost) all intended purposes, the training set error (MPE here) does not represent generalisable insights. Therefore I would suggest you focus on the validation and/or a separate test set. $\endgroup$ – usεr11852 says Reinstate Monic Nov 10 '19 at 1:51

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.