search

khliland / pls

The pls R package
36 stars 3 forks source link

Repeated k-fold CV #4

Closed dominix88 closed 5 years ago

dominix88 commented 5 years ago

Hi, thanks for the great package. Is there a way to run repeated k-fold CV to get a standard error for the prediction loss?

bhmevik commented 5 years ago

Estimation of stderr for prediction loss by repeated k-fold CV is not implemented, no. However, the selectNcomp() function to select optimal number of components estimates the stderr from the cross-validation residuals when called with method = "onesigma". Perhaps you can use that. See ?selectNcomp (especially the Details section). You can also look at the code of selectNcomp to see how the stderr is calculated.