Is there a way to do a stratified train/test split without shuffling the data?

Question

I'm using time sensitive data and would like to maintain the order of the data but stratifying the data since I've got multiple labels. I haven't found any libraries that allow this.

Answer 1

Hi Juanro could you provide with an example what are you trying to do as it might help to understand the problem in a better way Thanks:

Answer 2

Please refer to the train_test_split documentation.

You can do somthing like this:

x_train, x_test, y_train, y_test = train_test_split(x, y, test_size=0.33, random_state=0, 
                                                    stratify=y)

The stratify = y will give a stratified split with same proportions of class labels as the input dataset.