Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

RE: st: Hierarchical divisive clustering with Stata?


From   khigbee@stata.com
To   statalist@hsphsun2.harvard.edu
Subject   RE: st: Hierarchical divisive clustering with Stata?
Date   Mon, 08 Nov 2004 09:12:25 -0600

Herve Stolowy <stolowy@hec.fr> asks:

> To my knowledge, the cluster commands of Stata perform a
> hierarchical agglomerative clustering.
> 
> I would like to perform a hierarchical divisive (or descending)
> clustering. Do you know if there is special command for that in
> Stata?

Stata's -cluster- command does not include any divisive
hierarchical clustering routines.  The reason is that for all but
very small datasets (small N) the computational time of divisive
methods is unreasonable.

The "[CL] Cluster Analysis" manual mentions this and points
out that while algomerative methods have a large task of
checking

    N(N-1)/2

possible first combinations of observations, divisive methods
have the almost impossible task of checking

    2^(N-1) - 1

possible first divisions of the data.  For aglomerative methods
the computational burden grows quadratically with N.  For
divisive methods it grows exponentially.

The formulas for N above are just for the first aglomeration or
division.  Each additional step of clustering follows along
similarly.

Ken Higbee    khigbee@stata.com
StataCorp     1-800-STATAPC

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index