Here is a tutorial on interpretable decomposition of the forecasted signal @Afiq Johari https://nixtla.github.io/neuralforecast/examples/signal_decomposition.html

@Kin Gtz. Olivares Thanks for sharing this breakdown method, which seems really helpful. Right now, what I'm most interested in is figuring out how each of the exogenous variable affects NHITS' predictions. Can you help me understand that better?

Hi, I am also very interested in seasonal decomposition. Thank you very much for sharing! Sadly, the website for examples is down as for now, but I am would definitely love to learn more about this topic!

Hey @Jonathan Shang-Wen Chang https://github.com/Nixtla/neuralforecast/blob/main/neuralforecast/models/nbeatsx.py#L83 You can use the Exogenous basis, to decompose with the exogenous features too

@Kin Gtz. Olivares is there a tutorial available for this decomposition involving exogenous features? The closest resource I found is an excerpt from this article: https://www.sciencedirect.com/science/article/pii/S0169207022000413#sec4 My goal is to simulate various forecast scenarios of exogenous features and observe their impact on the target variable. Naturally, this assumes causal relationships between the exogenous features and the target variable.