How to use 'crossval' without pretrained model?

How can you generate a partitioned model using the 'crossval' function without creating a pretrained model first?

Answer 1

[MathWorks Support Team]

This functionality is not supported by the 'crossval' function. However, one workaround is to use a fit function with the 'CrossVal' name-value argument specified.

For example, to directly generate a partitioned SVM model, you can run the following command:

CV_SVM_model = fitcsvm(X,Y,'CrossVal','on')

To read more about the cross-validation options for the ‘fitcsvm’ function, please refer to the following documentation page: 

https://www.mathworks.com/help/releases/R2022a/stats/fitcsvm.html#d123e359846

Similar steps can be taken with fit functions for other model types.