[R] Questions about regression with time-series

[Giovanni Petris]

I think the function 'arima' can handle this, except for
the automatic evaluation of the ARMA orders.

Giovanni

> Date: Wed, 06 Dec 2006 12:06:46 +0100
> From: Lulla OPATOWSKI <lopatows at pasteur.fr>
> Sender: r-help-bounces at stat.math.ethz.ch
> Precedence: list
> 
> --===============0156932283==
> Content-Disposition: inline
> Content-Type: text/plain
> Content-Transfer-Encoding: quoted-printable
> Content-length: 954
> 
> Hi,
> 
> I am using 2 times series and I want to carry out a regression of Seri1=20
> by Serie2 using structured (autocorrelated) errors.
> (Equivalent to the autoreg function in SAS)
> 
> I found the function gls (package nlme) and I made:
> 
> gls_mens<-gls(mening_s_des~dataATB, correlation =3D corAR1())
> 
> My problem is that I don=92t want a AR(1) structure but ARMA(n,p) but the=
> =20
> execution fails :
> 
> gls_mens<-gls(mening_s_des~dataATB, correlation =3D corARMA(p=3D52))
> Error in "coef<-.corARMA"(`*tmp*`, value =3D c(11.2591629857661,=20
> 9.1821585359071,  :
> 	Coefficient matrix not invertible
> 
> This should be because most of the coefficients <52 are near to 0.
> 
> I am looking for a way to be able :
> - To evaluate automatically my ARMA structure (if it exists)
> - To specify manually the not null lags for my ARMA structure (as a=20
> vector for example)
> 
> Does anyone know about such functions?
> 
> Thank you for your help
> 
> 
> 	[[alternative text/enriched version deleted]]
> 
> 
> --===============0156932283==
> Content-Type: text/plain; charset="us-ascii"
> MIME-Version: 1.0
> Content-Transfer-Encoding: 7bit
> Content-Disposition: inline
> 
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
> 
> --===============0156932283==--
> 
>