should I separate the data as training and testing when I use cross validation in ensemble methods? and is there any fully useful code? Thanks much.

should I separate the data as training and testing when I use cross validation in ensemble methods? and is there any fully useful code? Thanks much.

Answer 1

[Prantik Chatterjee]

I am assuming that you have access to the latest MATLAB release.

You do not need to separate the data as training and test if you use the matlab library function 'crossval' for cross-validation. 'crossval' will, by default, split the given dataset into 10 folds internally. Alternatively, you can specify the number of folds using the argument 'Kfold' with 'crossval'. Please refer to the following documentation for more details on 'crossval'

https://in.mathworks.com/help/stats/classificationsvm.crossval.html

For a working example on using cross validation in ensemble methods, please refer to the following page

https://in.mathworks.com/help/stats/classreg.learning.partition.classificationpartitionedensemble.html