R help - May 2008 - ARIMA, AR, STEP

Here is my problem:
Autoregressive models are very interesting in forecasting consumptions (eg
water, gas etc).

Generally time series of this type have a long history with relatively simple
patterns and can be useful to add external regressors for calendar events
(holydays, vacations etc).

arima() is a very powerful function but kalman filter is very slow (and I foun
difficulties of estimation) while ar() is too simple but fast (but do not have a
method for forecasting I think)

Is there something like arima() but entirely implemented in C and efficient like
ar() ???
Is there something like step() for ARIMAX? It would be very useful for external
regressors.

Try the code below (imagine daily data for some years):

x <- rep(c(15,20,20,20,20,12,10), 5*52)
set.seed(1234)
x <- x + rnorm(length(x))

#plot(as.ts(x[1:21]))

#slow
arima(x, c(1,0,1), list(order = c(2,0,0), period = 7))
arima(x, c(2,0,0), list(order = c(3,0,0), period = 7))
#slower
arima(x, c(2,0,1), list(order = c(3,0,0), period = 7))
# do not converge
arima(x, c(2,0,0), list(order = c(3,0,1), period = 7))

#fast but not enough sophisticated
ar(x)

Thanks in advance
Daniele



________________________________
ORS Srl

Via Agostino Morando 1/3 12060 Roddi (Cn) - Italy
Tel. +39 0173 620211
Fax. +39 0173 620299 / +39 0173 433111
Web Site www.ors.it

------------------------------------------------------------------------------------------------------------------------
Qualsiasi utilizzo non autorizzato del presente messaggio e dei suoi allegati è
vietato e potrebbe costituire reato.
Se lei avesse ricevuto erroneamente questo messaggio, Le saremmo grati se
provvedesse alla distruzione dello stesso
e degli eventuali allegati.
Opinioni, conclusioni o altre informazioni riportate nella e-mail, che non siano
relative alle attività e/o
alla missione aziendale di O.R.S. Srl si intendono non attribuibili alla società
stessa, né la impegnano in alcun modo.

	[[alternative HTML version deleted]]

On Thu, 8 May 2008, Daniele Amberti wrote:
> Here is my problem:
> Autoregressive models are very interesting in forecasting consumptions (eg
water, gas etc).
>
> Generally time series of this type have a long history with relatively
simple patterns and can be useful to add external regressors for calendar events
(holydays, vacations etc).
>
> arima() is a very powerful function but kalman filter is very slow (and I
foun difficulties of estimation) while ar() is too simple but fast (but do not
have a method for forecasting I think)
>
> Is there something like arima() but entirely implemented in C and efficient
like ar() ???
You mean, like arima0()?

I am not sure arima() is inefficient, rather that you are asking for the 
solution to a computationally difficult problem (which in your example is 
looking to estimate structure that is not there!).
> Is there something like step() for ARIMAX? It would be very useful for
external regressors.
>
> Try the code below (imagine daily data for some years):
>
> x <- rep(c(15,20,20,20,20,12,10), 5*52)
> set.seed(1234)
> x <- x + rnorm(length(x))
>
> #plot(as.ts(x[1:21]))
>
> #slow
> arima(x, c(1,0,1), list(order = c(2,0,0), period = 7))
> arima(x, c(2,0,0), list(order = c(3,0,0), period = 7))
> #slower
> arima(x, c(2,0,1), list(order = c(3,0,0), period = 7))
> # do not converge
> arima(x, c(2,0,0), list(order = c(3,0,1), period = 7))
>
> #fast but not enough sophisticated
> ar(x)
>
> Thanks in advance
> Daniele
>
>
>
> ________________________________
> ORS Srl
>
> Via Agostino Morando 1/3 12060 Roddi (Cn) - Italy
> Tel. +39 0173 620211
> Fax. +39 0173 620299 / +39 0173 433111
> Web Site www.ors.it
>
>
------------------------------------------------------------------------------------------------------------------------
> Qualsiasi utilizzo non autorizzato del presente messaggio e dei suoi
allegati ?? vietato e potrebbe costituire reato.
> Se lei avesse ricevuto erroneamente questo messaggio, Le saremmo grati se
provvedesse alla distruzione dello stesso
> e degli eventuali allegati.
> Opinioni, conclusioni o altre informazioni riportate nella e-mail, che non
siano relative alle attivit?? e/o
> alla missione aziendale di O.R.S. Srl si intendono non attribuibili alla
societ?? stessa, n?? la impegnano in alcun modo.
>
> 	[[alternative HTML version deleted]]
>
>
-- 
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