0

I would like to run a cross-validation function like cv.glm on a logistic regression model built with speedglm on a large (millions of rows) data set. Does any such function exist?

I am finding that cv.glm (from boot package) and the train function from caret do not recognize speedglm models.

Thanks in advance.

Steven Barkin
  • 99
  • 1
  • 2
  • 8
  • You could build your own caret model. Or ask on github to add speedglm to the caret library. – phiver Nov 14 '16 at 08:32
  • Why do you need to cross-validate with that much data? Set aside 250K and use that to validate your model. Unless you have a really low event rate, that should suffice – topepo Nov 15 '16 at 02:19
  • Thanks, I will try to follow both of your suggestions. – Steven Barkin Nov 19 '16 at 01:30
  • You can try the `sperrorest` package. It provides a generic CV interface. All you need is a working `predict()` function. – pat-s Dec 23 '16 at 21:36

0 Answers0