> > > At 15:36 24/10/02 +0200, Christian Hennig wrote:
> > >
> > > >2) On clustering with R1=R2=R3=R. kmeans clustering implicitly assumes
> > > > clusters to have unit matrix correlation. So transforming the data
> to
> > > > unit covariance and then applying 3means will give clusters with
> > > > approximately R1=R2=R3=R.
> > >
> > > R1=R2=R3, maybe but =R???
...we look for clusters with unit correlation in data with unit correlation...
> > >
> > > Surely it is most unlikely that the overall correlation structure
> > > would mirror
> > > the withincluster structure? It is also hard to think why that might be
> > > desirable. If it were then an obvious way to achieve it would be
> > > to randomly
> > > allocate the data points to the three clusters.
> > >
> > > Murray Jorgensen
OK, my suggestion will usually not give perfectly R1=R2=R3=R and not a very
good clustering, but to a certain amount it tries to do both
simultaneously.
> I apologise if the matter is not strictly related with cluster analysis.
> Maybe it can be considered an allocationoptimization problem.
>
> Kind regards
> max
However, clustering does not seem to be in any sense Max'
aim (not only "not strictly"), and therefore all this does not really help.
I guess that under usual conditions my suggestion will match R1=R2=R3=R
worse then random allocation.
Christian

***********************************************************************
Christian Hennig
Seminar fuer Statistik, ETHZentrum (LEO), CH8092 Zuerich (current)
and Fachbereich MathematikSPST/ZMS, Universitaet Hamburg
[log in to unmask], http://stat.ethz.ch/~hennig/
[log in to unmask], http://www.math.unihamburg.de/home/hennig/
#######################################################################
ich empfehle www.boag.de
