Prediction on test set!! High Priority

[AnjaliSetiya]

Hi! I was reading research papers that have used mlxtend StackingCVClasssifier for stacking purpose. When these papers report prediction on test set they say that, In the testing set, five-fold CV model is used to predict the original testing set, again obtaining five predictions. In order to ensure the slitting ratio between the training set and testing set, so here the predictions are averaged horizontally to obtain a one-dimensional matrix.

My question is does StackingCV classifier handles this when we predict on test set, i.e. whether test is predicted across the folds and an average is taken. If so why is not mentioned anywhere in documentation. If not what is actually happening.

Describe the documentation issue

Suggest a potential improvement or addition

This will help in better understanding of the package