Imbalance Dataset _ KFold interest vs RandomUnderSampler

Module 7. Evaluating model performance Choice of cross-validation
#1

Hello,

Probably some of last of my questions (I know I am still at basic level…)

I am slightly confused about the use of KFold strategy in case of imbalanced dataset.

If I use a pipeline including for example randomundersampling from imblearn, do I need to use stratified Kflod solution? What is advantage of both?

Thanks always.

#2

To make the discussion easier to follow could you please edit your post and its title and only ask the first question here. Then open 2 other questions in 2 other posts.