Hello,
I have a paper accepted with revision at a solid journal. One of reviewers is insisting to do 10-fold CV to validate the proxy modeling. However, the paper talks about using three new algorithms to be used as proxy models.
Random subsampling cross validation has been used to test the efficiency of the new algorithms with two different objective functions.
The algorithms have been conducted in R and couldn't use 10folds CV with these algorithms. The available R package are mostly used with linear regression.
What should I do with this situation?
Should I ask the editor to ask another reviewer to judge in this specific point?
I should mention that I have never seen any paper used 10-folds CV to validate proxy modeling.
I know Junko worked on R too, he might be able to help you
Junko Hutahaean
Best Regards
Hi Watheq - well done for the paper.
Is there any particular reason you could not do the k-fold CV in R whereas you can do the random subsampling CV? Is that because the technicality in R implementation?
Junko
Thanks Zainab.
Yes Junko. Actually I did 10 folds CV with linear and robust linear regression. However, the other proxy models are implemented through a built-in packages, in which I cannot link with the Cv packages.
Hi Watheq - in that case I think you should code up your own k-fold CV function and link to the proxy built-in package. For this reason, I don't think the reviewer will take the technicality reason so not to use the k-fold CV.
- Badges
- Science topic
- Similar topics
- Social Science
- Media
- Journalism