# [R] correlation coefficient of ARIMA() or GLS() ?

From: Jan Verbesselt (Jan.Verbesselt@agr.kuleuven.ac.be)
Date: Tue 25 May 2004 - 22:52:33 EST

```Message-id: <000401c44257\$25f52fb0\$1145210a@agr.ad10.intern.kuleuven.ac.be>

```

Hi R-helpers,

I fitted the following ARIMA model onto de-seasonlised time series 1 and
2, which were strongly seasonal. How can the Rē or the coefficient of
determination for the structural term be calculated between these two
fitted time series?

Is an GLS() the solution (difference between ARIMA and GLS) ? Is it
possible to calculated an Rē from a GLS() model ? # nlme and MASS
package

***************************
reg.model <- arima(serie2.dd, order=c(6,0,0), xreg = serie1.dd, method =
"ML")
tsdiag(reg.model) # =>looks OK. residuals are not autocorrelated
(ljung-box statistic) but not normally distributed.Problem?

> reg.model
Call:
arima(x = serie2.dd, order = c(6, 0, 0), xreg = serie1.dd, method =
"ML")
Coefficients:
ar1 ar2 ar3 ar4 ar5 ar6 intercept serie1.dd

-0.1033 -0.3957 -0.0423 -0.2719 -0.2969 -0.1401 -2.3535 585.9007
s.e. 0.0843 0.0823 0.0831 0.0857 0.0795 0.0829 3.1520 118.1323

sigma^2 estimated as 6932: log likelihood = -829.9, aic = 1677.79
>

gls.ddm2 <- gls(serie2.ddm ~ serie1.ddm -1, correlation=corARMA(p=6),
method="ML") # fit of a gls without intercept because the intercept was
not significant but residuals are still auto correlated...?!
****************************************************

Tips, advice or examples are mostly welcome.
Regards,
Jan

_____________________________________________________________________
Jan Verbesselt
Research Associate
Lab of Geomatics and Forest Engineering K.U. Leuven
Vital Decosterstraat 102. B-3000 Leuven Belgium
Tel:+32-16-329750 Fax: +32-16-329760
http://gloveg.kuleuven.ac.be/

______________________________________________
R-help@stat.math.ethz.ch mailing list
https://www.stat.math.ethz.ch/mailman/listinfo/r-help