Jim,
I think you should get in touch with David Dubin about this. He and Mark
Rorvig presented a paper (somewhat at my instigation) on extension of the
"classical" (two-way) metric MDS method (most closely associated with Warren
Torgerson) to large data matrices, at a Workshop on MDS Algorithms organized
by Phipps Arabie, Larry Hubert and me, sponsored by DIMACS (directed by Fred
Roberts and co-directed by Mel Janowitz) at Rutgers, in August 2001. I believe
they had coorganized a session dealing with use of this and other methodology
for information retrieval at the Wisconsin meeting of CSNA last summer, but,
tragically, we were all shocked to learn that Rorvig died as a result of an
apparently longstanding heart problem, so that this became instead a memorial
session in Rorvig's honor. (I was not able to attend that CSNA meeting
myself,
so I am not sure if you were there or not.) I believe, however, that, despite
Mark Rorvig's untimely and unexpected death, some software was developed by
the
two of them that would deal with your problem of an MDS analysis of a 400x400
symmetric proximity (similarity or dissimilarity) matrix. Dubin may also have
carried this work further (and/or Rorvig and he may have prior to Rorvig's
death), so I definitely think this would be your most promising approach at
this point to dealing with this problem.
As for the version of an MDS program available in SAS (or in many versions of
SPSS-- although later professional versions of SPSS now include Willem
Heiser's
PROXSCAL, which is a very acceptable option), I would strongly recommend
against
using that if at all possible, even if it WOULD deal with such a large matrix.
SAS (and the student, and other-- earlier-- versions, of SPSS) use ALSCAL
for MDS
analysis, and there are numerous problems with that methodology that I think
would argue against using it-- particularly for THREE-WAY MDS analysis (but
also,
to a lesser extent for two-way MDS also ). If you want more information on my
reasons for recommending against using ALSCAL for MDS analysis, let me know and
I will send you such further info.
Best regards,
Doug.
At 01:25 PM 11/12/2002 -0500, F. James Rohlf wrote:
>The NTSYSpc software (http://www.exetersoftware.com/cat/ntsyspc.html) can
>presently go up to n = 362 for non-metric multidimensional scaling. That
>limit could be changed if necessary.
>-----------------------
>F. James Rohlf
>State University of New York, Stony Brook, NY 11794-5245
>www: http://life.bio.sunysb.edu/ee/rohlf
>
> > -----Original Message-----
> > From: Classification, clustering, and phylogeny estimation
> > [mailto:[log in to unmask]]On Behalf Of Dirk Meurer
> > Sent: Tuesday, November 12, 2002 12:01 PM
> > To: [log in to unmask]
> > Subject: MDS with a 400X400 Matrix
> >
> >
> > Dear Listmembers,
> > I would like to do a MDS with a 400X400 square, symetric matrix of
> > (dis)similaritys. Most MDS-Software is limited to a much smaller amount
> > of variables though. I have been told, that SAS might be able to process
> > the analyis I need, but this is quite inconvenient for me in terms of
> > access to the software, hardware-requirements etc. Could anybody tell
> > me, if there is a stand-alone program that can do MDS (preferably
> > nonmetric) with a matrix of that size?
> >
> > Thanks a lot for your help,
> > dirk
> >
> > P.S: My Matrix is not suitable for factor analysis and clustering does
> > not produce the results I need.
> >
######################################################################
# J. Douglas Carroll, Board of Governors Professor of Management and #
#Psychology, Rutgers University, Graduate School of Management, #
#Marketing Dept., MEC125, 111 Washington Street, Newark, New Jersey #
#07102-3027. Tel.: (973) 353-5814, Fax: (973) 353-5376. #
# Home: 14 Forest Drive, Warren, New Jersey 07059-5802. #
# Home Phone: (908) 753-6441 or 753-1620, Home Fax: (908) 757-1086. #
# E-mail: [log in to unmask] #
######################################################################
|
