On Sat, 8 May 2004 03:27:17 +0800, "Brian" <[EMAIL PROTECTED]> wrote:
> I would like to cluster a set of data. However, I don't know which > clustering method is the best. Would you please suggest a way to decide > which clustering method is the best? > What data? Since you don't know anything about clustering, you should probably consider reducing your data eyeball after looking at crosstabs, if there are a few disparate measures; or by factor analysis, if there are a large number of similarly scaled items. Here is an opinion, which I wonder if there is much objection to -- Clustering by computer is a moderately useless enterprise even in moderately skilled hands. Here is an opinion that I think is largely shared -- If you want to learn Clustering from scratch, you need to run one or two dozen analyses to see what happens, especially for data like your own, just to get started. That is true to a lesser extent for factor analysis, too. The SAS manual has a good section on clustering, written by Warren Sarle. Some of it is included in my own stats-FAQ, and (IIRC) a major part of it used to be online elsewhere. -- Rich Ulrich, [EMAIL PROTECTED] http://www.pitt.edu/~wpilib/index.html . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
