I would like to use one-standard -deviation rule for my hyper-parameter selection for elastic net. I understand how to do it for ridge or lasso, but when it comes to two regularization parameters I am confused. Is there any way of determining which model is more simple compared to the other one? (Because there could be several model within 1std of the minimum but have the same number of active sets). And I can't find any paper/book describing how to rigorously do this.

Here is the explanation of the one-standard-rule for cross-validation.


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.