On Extracting the Coefficient Vectors from NBEATSx Interpretable Configuration Model

[andrewrgarcia]

Description

Hello,

I'm currently working with the NBEATSx model as detailed in the paper "Neural basis expansion analysis with exogenous variables: Forecasting electricity prices with NBEATSx". On page 7 of the paper, it is mentioned that the forecast can be decomposed into trend, seasonality, and exogenous components, each with their model-adjustable $\theta^i_{s,b}$ coefficient vectors. This decomposition aspect is crucial for the interpretability of the model outputs.

While I understand that the forecast can be decomposed into separate components (as also illustrated in the Nixtla documentation), my primary interest lies in extracting the $\theta^i_{s,b}$ coefficients themselves.

Could you provide guidance on how to access or extract these coefficient vectors from the NBEATSx model? If direct extraction is not straightforward, could you point me to the relevant section of the source code where I might implement this feature or make necessary modifications?

Any assistance or direction you could provide would be greatly appreciated.

Thank you for your time and help.
-Andrew Garcia

Use case

No response