Hi all!

I need to split my dataset into 10 folds to do cross-validation "manually" (I know about the crossfold command, but it won't work for what I need).

I need a loop that allows me to run 10 cycles. In each of them, I need to select the train set (9 parts) and the validation dataset (1 part). Then I will run my model is the trainset and use the predict command to estimate the prediction of this model on the validation dataset.

This is what I have come up with so far, which I am aware it's wrong...

sysuse auto, clear

generate prp=0

* ** variable with the fold number. I know that in R there is a command that allows me to split n observations into K groups to be used for (repeated) K-fold cross-validation (cvfolds). Any suggestion?

egen split = seq(), f(1) t(10)

forvalues i = 1/10{

reg price mpg headroom if split != `i'

predict p if split =`i'

replace prp=p

drop p

}

Thanks a lot in advance!