Add support for fractions for window length parameters

Is your feature request related to a problem? Please describe. Currently, window length parameters (e.g. window_length or initial_window in SlidingWindowSplitter) have to be integers and denote the absolute number of time points, similar to how scikit-learn handles fractions in their cross-validation functionality.

Describe the solution you’d like Add support for floats denoting the number of time points relative to the passed data. For example passing window_length=0.5 would mean half of the data y passed to split in the SlidingWindowSplitter.

The parameters window_length and initial_window are used in several locations in the code base.

[Ifeanyi30]

I would love to

On Thu, Apr 15, 2021, 3:58 PM Markus Löning @.***> wrote:

@thayeylolu https://github.com/thayeylolu great! Opening a draft PR is usually a good first step to indicate that you’re working on it and to receive early feedback. @Ifeanyi30 https://github.com/Ifeanyi30 perhaps you’d be interested in reviewing once the PR is up?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/alan-turing-institute/sktime/issues/795#issuecomment-820496817, or unsubscribe https://github.com/notifications/unsubscribe-auth/AL45AALUE5FTT7EDWEKN5PTTI35HTANCNFSM42WU5AFA .

[thayeylolu]

@Ifeanyi30 Thank you for showing interest in this. @mloning I have started working on it