time series analysis

Hello out there,

in their "R Complements" to Modern Applied Statistics W. Venables and
B. Ripley say that there is a lack of time series related functions in
R compared to S plus. Looking at CRAN I didn't find much, too. As I
have to learn something about +-basic ts analysis, e.g. spectral
analysis, I would like to know whether somebody has developed a
"unpublished" time series package for analysis and plotting, which
he/she could make available. Unfortunately, my mathematical and
numerical background knowledge isn't sufficient to do it myself :-(

If there is a S library which could be used, I would appreciate hints,
too!

Many thanks for your help,
Ulf

I have some R functions that I use which you are welcome to try. They are R
programs as I float between Unix Macs and PC's. A word of caution, I am not
an ace programmer and they are not very efficient or good at error
trapping. As they are in R the maximum likelihood estimation is slow. If
you want to try them I will dig them out.

You might be better of with the Timeslab add on. These are the routines
from Newton's Timeslab book ( execellent) and are pretty good. I loaded the
Splus code from Statlib using dynamic load but  I understand that they are
now at CRAN as a package. They work and the documentation ( or at least the
stuff included at Statlib  in Tex (not Latex))is good.




  G.Janacek				          e-mail G.Janacek at uea.ac.uk
  School of Mathematics			          tel 44-(0)1603-592577
  UEA, Norwich NR4 7TJ,UK		          fax: 44-(0)1603-259515


-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
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
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

My time series package "DSE" is available and works just fine under R. It
provides considerable time series functionality which is not available in Splus.
(It does not duplicate everything that is in Splus. My interest has always been
in multivariate time series, and that functionality has historically been fairly
limited in Splus.) A user guide and the html help are available on the web at
<http://www.bank-banque-canada.ca/pgilbert>, so you can get some idea if it does
what you want before you install it.

There is a version of DSE in the devel area of CRAN and a slightly more recent
version available at my web site. It is in the devel area because the R package
install procedures were in a state of flux at the time I submitted it, not
because the code is unstable. The version on my web site dse.v.98.12alphaR
installs with R 0.63.2 which I am now using, and also with 0.63.1. As I recall,
the version on CRAN installed with 0.62.x, but probably does not install
correctly with 0.63.x.

Some of the documentation is still specific to S, but in most respects that
should not cause any problems. The versions on my web site are named "alpha"
because there are many new features which have not yet been properly documented.
The code is actually more stable than versions I have previously called "beta".
(In my experience something better than beta does not exist, even from
commercial vendors, only the label is occasionally removed.)

I work in Unix and have not tried to port the code to Windows, though if anyone
is interested I am willing to provide any information that I can. The main
problem is a fortran routine which is fairly critical.

In addition to DSE  the package CODA on CRAN has the function spec.pgram. (DSE
does not have much in the way of spectral analysis.)

Paul Gilbert

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
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
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._