Or, rather, it confirms my expectation that #354 is now fixed.
The model is agnostic of the timestep resolutions of the input data. It will just have NaNs where timesteps don't match.
Resampling then squashes all those NaNs out, so long as the resampling frequency is <= any of the input resolutions.
Fixes #354
Or, rather, it confirms my expectation that #354 is now fixed.
The model is agnostic of the timestep resolutions of the input data. It will just have NaNs where timesteps don't match. Resampling then squashes all those NaNs out, so long as the resampling frequency is <= any of the input resolutions.
Reviewer checklist: