Question

Can someone please explain to me the difference and its effect on partitioning data using either the stratified 10-fold cross validation procedure (as signified by ‘k’ partition type) or a 10-fold cross validation procedure (as indicated by “kfold” partition type).

My intention is to use this partitioning to perform a cross-validation procedure for time series analysis using a bootstrap ensemble. In this case, which partition type would be preferred to partition my explanatory variable to fit my response variable as best as possible?

Any help is appreciated.

Thank you

Answer 1

There is absolutely no difference. The cvpartition class constructor just checks the n characters you’ve provided against the first n characters of the standard method names. So 'k', 'kfo', and 'kfold' are all equivalent arguments.