arima0() (PR#314)

Prof Brian D Ripley
Sun, 7 Nov 1999 18:07:47 +0000 (GMT Standard Time)

On Sun, 7 Nov 1999 wrote:

> Full_Name: Ahmad Abu Hammour
> Version: rw0651
> OS: windows 95
> Submission from: (NULL) (
> Although I know that "ts package" is preliminary, I wanted to compare the
> results from R and SPSS. I ran ARIMA(2,1,2) in both softwares. I got NaN in
> standard errors of coefficients from R and real figures from SPSS. I changed
> "delta" in R to match that used by SPSS, I received results with no NaNs but

Can you please submit a full and reproducible example, as the FAQ suggests?

Does SPSS in fact do maximum likelihood estimation?  Last time I looked it
used an approximation. (You do need to be careful: for example S-PLUS's
arima.mle does not do MLE, despite its name.)  One isssue here is that most
packages do not ensure that the returned coefficients are valid for a
stationary invertible process: arima0 (by default) does.

The standard errors are part of the `preliminary' bit: they are waiting for
a better optimizer.  The `0' on arima0 indicates that it is intended to be
replaced fairly soon.

> different from those from SPSS specially in MA coefficients. They came out
> negative numbers while positive from SPSS. 

That is a matter of definition. The signs of each of the AR and MA
coefficients are often swapped, so I would not be at all surprised that
SPSS is different. R is intended to be consistent with S-PLUS (as
documented in Venables & Ripley, at least).

Brian D. Ripley,        
Professor of Applied Statistics,
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-devel mailing list -- Read
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: