hello, i might be missing something but is there a way to train NHITS on a timeseries with two target variables?
i.e. having a two-element 'y' target
c
Cristian (Nixtla)
10/07/2023, 10:52 PM
Hi @Tomáš Halama. Not currently, the NHITS is an univariate model. You can train a single global model that will forecast both target variables independently. You need to specify them in separate
unique_id
values
t
Tomáš Halama
10/08/2023, 2:09 PM
that doesn't work for my use case sadly, i have several thousands independent timeseries where each has two components 😕