Re: [Gretl-devel] Estimating forecast error
Am 13.12.2010 14:33, schrieb Gústaf Steingrímsson:
Hi
I am wondering wether it is possible to estimate forcast error in Gretl
in simple terms. Let‘s fx. say I have a time series on monthy basis and
the forecast model is simply some version of lagged values of this same
time series (ARIMA,AR etc.). now I want to estimate historical forecast
error for one, two, ..., n months in the future or get some data table
where one, two, ..., n months forecasts can be compared to real
historical values. Is this possible?
Hi, does the accessor '$fcast' after a 'fcast' command do what you want?
See the explanation for the fcast command.
(BTW, I'm noting that $fcast and $fcerr are missing in the documented
list in the function reference menu.)
HTH,
sven