Sorry meant to add more detail --

I have done some clustering for a marketing promotion and being not
especially stats based think it could have been more robust.  Having
done some research what I have been grasping for it seems is Chaid and
I have invented instead a childlike version of it which looks for
large variations against the norm then splits these out into 2 further
groups.  Becasue of the large number of possible permutations of using
multiple variables what I have come out with is some arbtrary groups
that fitted my original preconceptions of what the groups should look
like - not very scientific!  Anyone out there got any simple chaid
type code for the arts-degree'd SAS users amongst us ;-)

You can start with this search :

http://search02.sas.com/query.html?qt=chaid&style=sup&nh=40&qp=&qc=suppsas&ws=1&qm=1&st=1&lk=1&rf=0&oq=&rq=0



With http://www.google.fr/search?hl=fr&q=SAS+CHAID&meta=

I found www.stat.lsu.edu/faculty/moser/exst7037/treeiris.pdf


Cordialement/
Best regards,

Stephane COLAS

**************************
Societe Datametric

Notre site / Our site
http://www.datametric.fr
**************************


Selon matt_hutch2002@YAHOO.CO.UK:

> Sorry meant to add more detail --
>
> I have done some clustering for a marketing promotion and being not
> especially stats based think it could have been more robust.  Having
> done some research what I have been grasping for it seems is Chaid and
> I have invented instead a childlike version of it which looks for
> large variations against the norm then splits these out into 2 further
> groups.  Becasue of the large number of possible permutations of using
> multiple variables what I have come out with is some arbtrary groups
> that fitted my original preconceptions of what the groups should look
> like - not very scientific!  Anyone out there got any simple chaid
> type code for the arts-degree'd SAS users amongst us ;-)
>