[R] ARIMA0 with xreg,

Brian Scholl brianscholl1973 at yahoo.com
Wed Dec 19 22:47:16 CET 2001

Hi all,

Using ar(), I fit a VAR to my time series that has a
reasonably 'nice' error spectrum and aic determines
p=7.   But the output for ar isn't quite as convenient
as arima0, namely in that it takes me an extra step to
get the s.e.'s of parameters and it doesn't produce an
estimate of the log-likelihood for comparison to other
models. So I thought I'd use arimia0, with xreg=x, my
matrix of data and order=(7,0,0).  

When I do this, the results are funny.  The residuals
look - well, just like the original series on a
readjusted scale.  The spectrum for the residuals
looks slightly different from the spectrum of x and
bears no resemblance to the nice spectrum I got before
(this one looks like a long memory process, while the
earlier was closer to white noise).  Certainly I don't
expect the spectra and residuals to look exactly alike
because of the different solution methods, but these
bear little resemblance to each other.

I assume, of course that the error is on my part,
perhaps something I've overlooked, but I'm not able to
find it.  




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