I would appreciate getting references regarding the
reliability of hierarchical clustering when repeated random
samples are supplied to standard clustering methods such as
Ward's, Centroid, Complete linkage, ... .  My own experience
is that reliability is quite low, but I am not aware of
published results.

        I have been unsuccessful at requesting the OLD
archives of class-l from the WAIS server and was not given
permission to use INDEX CLASS-L by the old listserv. Is
there a way to use listserv at the old host to get them? If
so I would appreciate knowing if this topic has been
discussed, and when.

Thank you, in advance.


Brian Schott/Decision Sciences        (404) 651-2005    [log in to unmask]
1000 J Mack Robinson College of Business      /      Georgia State Univ
Atlanta, Georgia USA 30303-3083           interests: approx. reasoning,
http://www.Gsu.EDU/~dscbms/         (B=)       decision support systems