[R] ARMA and ARIMA modeling

Paul Gilbert pgilbert at bank-banque-canada.ca
Mon Mar 11 23:37:16 CET 2002

David Brahm wrote:

> Could some kind soul explain the relationship among packages "ts", "tseries",
> "dse", "dse2", and "fracdiff"?  Are they 'competing' products or does one
> depend on another? 

"dse2" uses "dse1" (they are in a bundled together) but mostly does not require
"ts", "tseries", or "fracdiff", although there are a few things where functions
from "ts" are used (eg. acf) and there is certainly the possibility users would
want to use all these libraries together, as I think they are mostly

> Where would be the best place for a novice to begin?

As usual, it depends what you want to do. If you are interested in multivariate
time series then dse is probably the place to start, and if not, then probably
not. Dse "out-of-the-bundle" does time-invariant, constant coefficient,
multivariate state-space and ARMA (VAR, VARX, ARIMA, ARIMAX), but not ARCH or
GARCH. There are tools for building other kinds of multivariate time series
models and tools for evaluating estimation techniques, but you would need to do
some programming. If you start with dse then start by reading the User's Guide
included in the bundle.

If you are interested in financial time series then dse "out-of-the-bundle" is
not great because you will probably want ARCH, GARCH, or some extension, so you
may be better off with other packages. However, I believe, if you are interested
in multivariate financial time series then you are going to need to do some
programming. In that case you may find dse a useful starting point.

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

More information about the R-help mailing list