At 15:44 +0200 07/11/2000, Kai Arzheimer wrote:
>Sorry, this is certainly a stupid question:
>
>Would make sense to apply some sort of rotation (like VARIMAX in PCA) to
>a multidimensional HOMALS-Solution? Any help, comments, pointers welcome
>
>Kai Arzheimer
>--
>Wissenschaftlicher Mitarbeiter
>Institut f¸r Politikwissenschaft
>Colonel-Kleinmann-Weg 2 / D-55099 Mainz
>Tel: 49-6131-39-23782 / Fax: 49-6131-39-22996
>* http://www.politik.uni-mainz.de/kai.arzheimer/Welcome.html
>* There are 3 kinds of people: those who can count & those who can't.
>
>
>=================================================================
>Instructions for joining and leaving this list and remarks about
>the problem of INAPPROPRIATE MESSAGES are available at
> http://jse.stat.ncsu.edu/
>=================================================================
Standard HOMALS practice is to look at one (for scale construction),
two (for mapping) or three (as an exception) dimensions. And then
what is interesting is usually the global shape of the point set. Not the
projections on arbitrary dimensions, which are only a usually
unsatisfactory tool to handle higher dimensionality. In
three dimensions dynamic rotation seems suitable. See
http://www.stat.ucla.edu/journals/jss/v01/i02/
http://www.stat.ucla.edu/journals/jss/v02/i08/
for nice programs to do this. See
http://baserv.uci.kun.nl/~rkonig/rotation_princals/
for a paper on rotation of PRINCALS (where it makes more
sense).
The notion of "dimensionality" in HOMALS is not as simple as
in ordinary PCA (or PRINCALS). See the attached pdf file.
--
===
Jan de Leeuw; Professor and Chair, UCLA Department of Statistics;
US mail: 8142 Math Sciences Bldg, Box 951554, Los Angeles, CA 90095-1554
phone (310)-825-9550; fax (310)-206-5658; email: [EMAIL PROTECTED]
http://www.stat.ucla.edu/~deleeuw and http://home1.gte.net/datamine/
============================================================================
No matter where you go, there you are. --- Buckaroo Banzai
http://webdev.stat.ucla.edu/sounds/nomatter.au
============================================================================
=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================