How can I incorporate features generated by `tsfresh`/`tsfeatures` when training an

NHITS

model? These generated features typically represent summarized values over the entire time period. Therefore, each

unique_id

would have a single value for each feature.

Hey. If they're constant over time you can specify them through

stat_exog_list

Here we have a tutorial on how to pass exogenous covariates, including both temporal and static: https://nixtla.mintlify.app/neuralforecast/examples/exogenous_variables.html