[R] ARIMA - h-step ahead errors

[Prof Brian Ripley]

On Sat, 18 Oct 2008, Nuno Prista wrote:

> Dear colleagues,
>
> “arima” returns directly the 1-step ahead errors but I am interested in 
> obtaining other h-step ahead errors for several ARIMA models I have fitted. 
> Is there any way I can obtain this with R? Any help would be appreciated.

See ?predict.Arima, the predict() method for its output.

Note that arima() returns the fittted innovations: these are not 
necessarily the '1-step ahead errors' but estimates of them.  (E.g. think 
about missing values.)

> Sincerely,
>
> Nuno Prista
> _________________________
> CO - FCUL, Lisboa, Portugal
> CQFE - ODU, Norfolk, USA


-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595