Approximating NHITS to NBEATSx Model:
Hi! I've been using NHITS to train my model, but I'm interested in understanding the effects of exogenous variables better.
I've heard that NBEATSx can provide such explanations. Could you please advise on how I can approximate my optimized NHITS model to a similar NBEATSx model, especially in terms of explaining the effects of exogenous variables?
Using SHAP and LIME for Model Explanations:
Now, assuming I manage to transform my model into an NBEATSx-like model with better explanation capabilities, I'd like to utilize SHAP and LIME for model explanations.
Could you guide me on the steps to use SHAP and LIME effectively with this NBEATSx-like model to gain insights into feature importance and predictions?