Date:         Thu, 11 May 2006 09:01:19 -0700
Reply-To:     "shunhung@gmail.com" <shunhung@GMAIL.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         "shunhung@gmail.com" <shunhung@GMAIL.COM>
Organization: http://groups.google.com
Subject:      Using Fastclus to score a new dataset
Comments: To: sas-l@uga.edu
Content-Type: text/plain; charset="iso-8859-1"

I used Fastclus to generate an output with 5 clusters using
Standardized variables (I have 10 variables). I want to use the cluster
centroids to score a new dataset.

A snippet of my code is below.

The output from outstat gives me standardized results. When I apply
this to score a new dataset, I am not sure how to read in the new
dataset. If I standardize the new dataset, the results get thrown off.
I can't get unstandardized data because some values have wider range
than others.

Has anyone experienced this problem before? Please help! :)


proc standard data=temp out=stand mean=0 std=1;
proc fastclus data=stand out=Clust maxclusters=5 maxiter=100
outstat=outcent;
run;


* "score" the new dataset using final seeds from above **;
proc standard data=temp2 out=pzipstd mean=0 std=1;
proc fastclus data=pzipstd instat=outcent out=scored;
run;