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
It's a common place to use features generated from timeseries itself like lags, rolling statistics and etc.
I've seen some approaches like using sklearn transformers and pipelines for fitting, and forecasting via recursive procedure( iteratively generating one step ahead predictions )
Have you any ideas about it?
I think it would be nice feature.
The text was updated successfully, but these errors were encountered:
Thanks for your inputs, we will put this on our radar! It might indeed be nice to have a subclass of ForecastingModel that can take any regression model to produce predictions using autoregression. In the meantime, if you're interested in contributing to Darts, please check out this post for guidelines to implement a new model. Also, please let us know if you have any other ideas!
Here are some elements of answers:
Some models are auto-regressive in nature (e.g AR(I)MA) and so this is already embedded in some models. Some other models such as neural networks are not (always) auto-regressive in nature but rather work by predicting fixed-length outputs. For these cases (neural networks), Darts now provides a native auto-regressive functionality, which will recursively call the neural net on its own outputs until predictions of a certain horizon length are obtained.
As auto-regressive forecasting is a somewhat "abstract" feature that is quite model-dependent, I'm closing this issue. Thanks for the comment @martins0n
Nice work, guys!
It's a common place to use features generated from timeseries itself like lags, rolling statistics and etc.
I've seen some approaches like using sklearn transformers and pipelines for fitting, and forecasting via recursive procedure( iteratively generating one step ahead predictions )
Have you any ideas about it?
I think it would be nice feature.
The text was updated successfully, but these errors were encountered: