[19267] in s-news-athena
[S] RE: Oswald in S-PLUS 2000 (2)
daemon@ATHENA.MIT.EDU (David Smith)
Mon Sep 20 06:48:57 1999
Message-Id: <5FD4320911EDD21194CB00C04F8EEB1B0877D6@tech_1.mathsoft.co.uk>
From: dsmith@mathsoft.co.uk (David Smith)
To: "S-News (E-mail)" <s-news@wubios.wustl.edu>
Cc: "'INSH (Ingrid Sofie Harbo)'" <INSH@NOVO.dk>
Date: Mon, 20 Sep 1999 11:40:42 +0100
Mime-Version: 1.0
Content-Type: text/plain;
charset="iso-8859-1"
I was asked recently whether any of the Oswald functions have been
incorporated into S-PLUS. As the Oswald author I'd have responded, but
Brian Ripley kindly answered before me.
Prof Ripley is correct. None of the functions in Oswald are incorporated
into S-PLUS 2000. The "variogram" function referred to is part of the new
Bates/Pinhero lme/nlme functionality in S-PLUS 2000, which incidentally can
fit a much richer class of models than Oswald ever could. I recommend NLME
heartily (speaking as a statistician with interest in LDA, not on behalf of
MathSoft), it's an excellent implementation. The (almost) latest version of
NLME is included in S-PLUS 2000, and you can get the very latest version
from the NLME website at
Ripley is also correct about GEE in Oswald. I haven't updated Oswald in
some time, and the GEE implementation from Vince Carey has since been
updated. The source for the latest GEE code is mentioned in the Oswald
readme file, but I reproduce it here for the list:
http://biosun1.harvard.edu/~carey/index.ssoft.html
Compiled versions for Windows (courtesy Brian Ripley) are available from
http://www.stats.ox.ac.uk/pub/SWin/.
I no longer maintain Oswald, partly because of lack of time in my new job,
but mostly because most of its application area is now better implemented by
Bates/Pinhero in their NLME package. It's still available from my Lancaster
website at
http://www.maths.lancs.ac.uk/Software/Oswald
and will continue to be so. (Although the mixed models implementation is
supersended by NLME, some may still be interested in the missing data
methods and the data type implementation.) NB: the version of Statlib is
now out of date.
# David Smith
--
David M Smith (dsmith@mathsoft.co.uk)
Mathsoft International, Bagshot, UK
Tel +44 01276 452299 x212
Fax +44 01276 451224
-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu. To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message: unsubscribe s-news
