I've been working with the NHITS library recently and I must say, it's incredibly well-designed and efficient. Great job on that!
I have a question about handling categorical variables within the NHITS model. Is there a built-in method for encoding categorical variables implicitly, or an option to specify that certain variables should not be scaled by the provided scaler type (when I manually provide the one hot vars)?
Thank you for your time and keep up the excellent work!
10/31/2023, 7:33 PM
Hi @Tomáš Halama! Thanks for using the library 🙂. We currently dont have a built-in method for encoding variables or scale only some features. However, scaling the one hot vars should not be an issue