Date:         Thu, 17 May 2007 20:25:28 -0700
Reply-To:     David L Cassell <davidlcassell@MSN.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         David L Cassell <davidlcassell@MSN.COM>
Subject:      Re: Clustering using SAS
In-Reply-To:  <200705171939.l4HI0o6u003205@malibu.cc.uga.edu>
Content-Type: text/plain; format=flowed

vuw100@YAHOO.COM wrote:
>
>Hi All,
>
>I have a question about clustering in SAS,
>
>I have 14 attributes, 25 observations, I want to cluster them, I dont have
>much preference about type of clustering (K-means, Hierarchial) etc.
>
>My only concern is that I want the clusters made so that 1st cluster has 5
>members and the second cluster the remaining 20 members,
>
>Can anyone please guide me whether this is possible and if so is it a good
>practice to do such a procedure?
>
>Thank you
>
>Vijay

I would say that this is not possible.

Now, if you have further information, such as "these represent 25 skeletons,
an unspecified set of 5 of whom have characteristics that should mark them
as hominid, while the others are not" that would help.  And it would explain
why you think you should get 5 and 20.

But without further information, I can't see how you can think that
your data would reasonably aggregate into sets of 5 and 20.

If your data have features which are clear, then *any* technique ought to
find them.   Although I would skip clustering methods like Ward's which
tend to try to lump the data into equal-sized clusters...

HTH,
David
--
David L. Cassell
mathematical statistician
Design Pathways
3115 NW Norwood Pl.
Corvallis OR 97330

_________________________________________________________________
More photos, more messages, more storage—get 2GB with Windows Live Hotmail.
http://imagine-windowslive.com/hotmail/?locale=en-us&ocid=TXT_TAGHM_migration_HM_mini_2G_0507