Closed fissehab closed 4 years ago
Unfortunately this is not possible. We use stats models under the hood and to the best of my efforts to read into this, they don't allow it. I found an interesting answer on stack exchange about this, but it's not something we'll support in the foreseeable future.
I am fitting regression with ARIMA errors using auto.arima. Since I suspect the models could be over-fit, I want to use regularization with it. Is it possible to implement lasso or elastic net regularization ?