Hi, thanks for the great package! I frequently work with many time series, and having had to write much of the same functionality in the past (i.e. fit many time series models in parallel), I appreciate how well it is done here.
Nonetheless, I was wondering what the differences between stream and refit really were - is one potentially more computationally efficient, or more stable, and do they perform the same operation?
From my current understanding, stream( model, new_data ) is roughly equivalent to refit( model, new_data, reestimate = TRUE ). Are there any subtle differences to worry about, and, if not, is there a reason to prefer one to the other?
JSzitas
commented
2 years ago
Hi, thanks for the great package! I frequently work with many time series, and having had to write much of the same functionality in the past (i.e. fit many time series models in parallel), I appreciate how well it is done here.
Nonetheless, I was wondering what the differences between stream and refit really were - is one potentially more computationally efficient, or more stable, and do they perform the same operation?
From my current understanding,
stream( model, new_data )is roughly equivalent torefit( model, new_data, reestimate = TRUE ). Are there any subtle differences to worry about, and, if not, is there a reason to prefer one to the other?Thank you!