-
-
Save yogabonito/b532e20b64587c19198f1aacb5ddfca2 to your computer and use it in GitHub Desktop.
Sorry for the late reply, @iTrooz! (I'm not that often on Github.)
I just had a look the code I wrote back then and I saw that I put my VECM code in its own direcotry (statsmodels/tsa/vecm/vecm
). It looks like the statsmodels team moved the VECM-related code to the vector_ar
directory after my code was merged into the statsmodels repository.
Very helpful. Thx!
I have a problem about forecast error variance decomposition (FEVD) that I didn't find relevant functions in statsmodels.tsa.vector_ar.vecm
. Is FEVD not supported yet?
Hi @RangerShaw,
I think the best way to ask is to open a new issue in the statsmodels repository. As I stopped working on statsmodels almost six years ago, I am not sure what features were added since then.
Deeply thank you for your demostrartion ~~~
Thanks for useful example
Awesome and thanks for you sharing! I have a question, the function select_order is to find the best lags of VAR model based on information criteria. But the k_ar_diff in function select_coint_rank is the lagged difference (VECM function as well). Should they be subtracted one from the result of select_order? (means k_ar_diff =2 in select_coint_rank and VECM, but not 3 which is the best order in the result of select_order)
I had a problem with
statsmodels.tsa.vecm.vecm
(ModuleNotFoundError) and had to replace it withstatsmodels.tsa.vector_ar.vecm
(so the line would befrom statsmodels.tsa.vector_ar.vecm import *
) to make it workAny idea why ?