Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, is already up and running.

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: Re: Clustermat puzzle

From (Brendan Halpin)
Subject   st: Re: Clustermat puzzle
Date   Sun, 25 Mar 2012 15:55:24 +0100

Following up to myself:
On Sat, Mar 24 2012, Brendan Halpin wrote:

> I have a small matrix of pairwise distances (all integers) that I'm
> passing to clustermat (Ward's method). I notice that if I scale the
> distances by a constant, I get different results. 

On mature reflection, this is clearly a precision issue. What must be
happenning is that at some stage in the clustering, when the candidates
for the next agglomeration are considered, what is a clear tie with the
integer distance matrix, results in a miniscule difference with the
scaled distance matrix. Using the integer matrix, the next agglomeration
is picked from among the ties (at random, or arbitrarily); using the
scaled matrix, the next agglomeration is determinately the one that has
just (illegitately) shaded the comparison.

Is there any way of controlling the tolerance used in the test for ties?
As far as I can see, the documentation for -cluster- and -clustermat-
doesn't discuss ties or tolerance. 


Brendan Halpin,   Department of Sociology,   University of Limerick,   Ireland
Tel: w +353-61-213147  f +353-61-202569  h +353-61-338562;  Room F1-009 x 3147    ULSociology on Facebook:         twitter:@ULSociology
*   For searches and help try:

© Copyright 1996–2015 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index