I\'m building a decision tree from scratch without the use of the sklibrary. The current method I\'ve used to split the sample is the K-fold method. I would like to switch from