This message was deleted.
# neural-forecast
s
This message was deleted.
c
Hi @Luis Huergo! Yes, you can do it with all models. Set the hyperparameter
exclude_insample_y=True
when you instantiate a model, and it will only use other variables to produce the forecasts
you still need to pass the
y
variable to the training data, because it will use it for the training loss.