Replies: 1 comment
-
Hi @hweiTPR, I've been using the hack outlined here #652 (comment) in the past to make GPVAR work with dynamic features. Maybe you can give it a try. |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
I'm trying to forecast for a small set of multivariate timeseries. They have dynamic features associated with some of them. I want to alter those dynamic features during predictions so that I can simulate forecasts under different future conditions.
I've tried
GPVAR
, and it doesn't seem to take any dynamic features right now. While adding those features directly as timeseries produced reasonable forecasts, it doesn't allow me to simulate forecasts under different future conditions.I also tried
DeepAR
which takes in dynamic features but didn't produce good forecasts. I suppose that's because it assumes univariate timeseries.Is there another model I can use that allows feat_dynamic_real. Or is there some relatively easy ways I can change the backend code to make this possible? I'm pretty new to this package, would appreciate someone points me to the right package or places.
Beta Was this translation helpful? Give feedback.
All reactions