[R] Arima

Prof Brian Ripley ripley at stats.ox.ac.uk
Sun Dec 16 22:45:11 CET 2001

On Sun, 16 Dec 2001, Pascal Grandeau wrote:

> I did a regression with ARMA errors using arima0 with
>    ari<-arima0(y,order=c(2,0,2),xreg=reg1,delta=-1)
> or
>    ari<-arima0(y,order=c(2,0,2),xreg=reg1)
> where reg1 is the matrix of the regressors and when I see diag(ari$var.coef)
> I get negative terms. Do you know what this mean ?

The optimizer failed to converge, probably. You did check the
convergence value in the fitted object, didn't you?

> I try to change transform.pars to 0 or 1 but this crash R on Windows.

Well, transform.pars=2 first runs transform.pars=1, so this is strange.
But probably your model is inappropriate.

> Is it possible to test the significativity of the estimators obtained by
> arima0 and how ?

Well, first you need to get it to converge. Then you can use Wald tests or
(better) likelihood ratio tests.  The log-likelihood is in the fitted

> I use arima0 because I have regressors and it seems it is impossible to uses
> arma() in tseries with regressors.
> Does anyone make a routine for regression with ARMA errors with least
> squares ?

What does that mean? `with least squares' implies independent errors.
arma() fits by so-called *conditional* least squares: that leaves out
terms in the log-likelihood which can be important, especially near
non-stationarity. I've never understood why anyone would want to do that,
except as a poor man's computational approximation.

> Another question : how can I handle missing values in regression with ARMA
> errors with R ?

As yet there are very limited possibilties: see e.g. na.contiguous.

You might want to investigate the dse bundle.

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 272860 (secr)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595

r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch

More information about the R-help mailing list