You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In order to force predictions to be positive (this is a common requirement for many business-related time series), the prophet package can be used with the ProphetPos class (see "Approach 5" here: facebook/prophet#1668 (comment)).
It would be nice to add a "ProphetPosModel" to Kats, so we can force predictions to be positive when we use prophet in Kats.
Maybe a positive constraint could also be added to the other models supported by Kats (SARIMA etc.).
The text was updated successfully, but these errors were encountered:
Thanks for the suggestion and we will keep this in mind for new feature development! Meanwhile, would it make sense to transform the data first (e.g., y' = log(ts)) and build model based on y' then transform it back. In this we can solve the problem of negative forecasts.
In order to force predictions to be positive (this is a common requirement for many business-related time series), the prophet package can be used with the ProphetPos class (see "Approach 5" here: facebook/prophet#1668 (comment)).
It would be nice to add a "ProphetPosModel" to Kats, so we can force predictions to be positive when we use prophet in Kats.
Maybe a positive constraint could also be added to the other models supported by Kats (SARIMA etc.).
The text was updated successfully, but these errors were encountered: