I using a mixed-effect logistic regression model and I need to do a cross-validation to check the accuracy of my model. I have 20 observations for each individual and 500 individuals in my dataframe.
Can I take, for example, 15 random observations of each indiviual and put this on my train set? And use the rest on the test set?
or can I put individuals that don't appear on the train set on the test set?