The question comes from when I was wondering about the cross-validation, finding out the best algorithm.

Then I got the question like if this model did a better job than the other, can the combination of two on different subpopulation be a better algorithm. Let say the model of KNN did a pretty job predict these types of people on these attributes and did a bad job on a complex outcome due to the large dimension. we use KNN for the subset that is small then the rest used another better method may be logistic regression.

  • $\begingroup$ Why should this be a problem? Assuming that you're not overfitting to the training data, why not? $\endgroup$ – Tim Jan 10 at 15:34
  • $\begingroup$ You are totally out the point, I did not ask to solve the problem, I ask if it could and may it be a better solution. $\endgroup$ – EconBoy Jan 10 at 15:46

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Browse other questions tagged or ask your own question.