Nikhil Gupta

From @ltsaprounis - https://robjhyndman.com/hyndsight/forecasting-within-limits/ Will add this transformer to sktime...

PR in sktime added: https://github.com/alan-turing-institute/sktime/pull/1913

Hi @AJarman , Yes that is correct. It has to be added as a preprocessing step in the pipeline which will appropriately transform the data to keep it within the...

Thank you. I appreciate your patience with this one.

@AJarman @ltsaprounis - Here is another option to limit the values but this uses clipping vs. transformation. I wonder which one is better? https://github.com/alan-turing-institute/sktime/issues/2291#issuecomment-1077868581

> just to clarify that I have that step to only make sure non-negative values are not fed into models that cannot handle that. I personally prefer having the "full"...

> If the user simply wants to clamp predictions they could do this themselves but the implementation suggests we want to fit/evaluate the forecasters based on user-defined clamps @AJarman By...

@atsangarides - Got it. In any case, the way this will be implemented, user will be in control. They can decide to clamp the predictions or just leave it as...

@atsangarides Would love for you to try out PyCaret Time Series (based on sktime) and provide feedback. If you would be open to it, please connect on LinkedIn and I...

@AJarman Could you submit a PR so we can review it (even one without tests would be ok for now until you add them later)? Thanks!