Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Forecasts with only Positive values #71

Closed
raedbsili1991 opened this issue Aug 10, 2023 · 1 comment
Closed

Forecasts with only Positive values #71

raedbsili1991 opened this issue Aug 10, 2023 · 1 comment

Comments

@raedbsili1991
Copy link

raedbsili1991 commented Aug 10, 2023

Another topic that may be worth talking about, add a function to the library, and potentially document it, is how to constraint the Forecasts.

For example, in my case, my dataset has always a target value y positives.
So I am looking how to constraint my forecaster to forecast only positive values. When trying to perform the automatic forecasts in function of a database, often I get models forecasting negative values, which doesn't make sense (think of y such as Pressure value or ... )
Does the library has a function that allow to do this ? One way is maybe add a Log transform then Exp reverter.

@mikekeith52
Copy link
Owner

The prophet model has something that allows you to define a model's "floor" when using logistic growth, but I'm not sure there is a good way to generally apply such an idea to all models. Artificially constraining point estimates can lead to unforeseen biases. Furthermore, applying transformations makes it hard to determine what values should be allowed to keep the final-evaluated points below or above a certain threshold. If you have a specific idea to modify the code to allow what you are asking for, I can take a look at it. Otherwise, this isn't something I will plan to work on.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants