[Gretl-users] Re: Recursive forecasts for Arima models
Am 24.11.2024 um 12:18 schrieb Artur T.:
Am 24.11.24 um 11:59 schrieb Sven Schreiber:
>
>
> fcast h --recursive # prints out many forecast values, each h-step
>
[but no values computed for the end of the possible range]
I think this part is buggy from the gretl side.
It certainly feels like it. I checked the C-level function
recursive_OLS_k_step_fcast in lib/src/forecast.c, but nothing obviously
wrong there, AFAICT.
Notice that the problem only appears for k>1, so maybe the k_step_init
function (which is called from recursive_OLS_k_step_fcast) has a problem.
cheers
sven