0

The models are trained using the same dataset.

I know these criterions can be use to do model selection for nested features. Suppose model 1 has feature A,B,C, model 2 has feature B,C,D, I can use these criterion for model comparison. But what if I have models of A, B, C, can I use BIC to compare to model with features D, E, F?

kjetil b halvorsen
  • 77,844
StaEx_G
  • 113
  • try https://stats.stackexchange.com/search?q=aic+non+nested ... (and similar searches). There's many posts that relate to this question – Glen_b Oct 25 '22 at 00:26
  • This doesn't quite seem to be a completely settled question, e.g. see https://stats.stackexchange.com/a/10165/805 ... there was a bit of back and forth in the literature, if I recall. – Glen_b Oct 25 '22 at 00:28

0 Answers0