Date:         Mon, 4 Nov 2002 14:44:08 -0500
Reply-To:     "Bhardwaj,Vatsal" <vb@EE.DUKE.EDU>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         "Bhardwaj,Vatsal" <vb@EE.DUKE.EDU>
Organization: Duke University, Durham, NC, USA
Subject:      Help with CLUSTERING procedures
Content-Type: TEXT/PLAIN; charset=US-ASCII

Hi Guys,

I am new to SAS and have just written my 2nd program. :-). I need some
help with clustering. I have a large Data set on which i am using FASTCLUS
to cluster data in to various states. The result files are very exhautive,
but unforutnately i have not been able to find what i need.

Is there anyway i can find out for every observation in my data to which
cluster that particular observation belongs to? . (After i have done the
clustering).

I have include my SAS program below. I tried lookign around the SAS Doc
but could not find anything helpful. Would appreciate any pointers on
this.

My measure of intereast here is to set up a n*n matrix( i have 'n' cluster states) for no. of
times transition occurs between any two clusters.

Thanks in Advance for your help and time.

                                              Best Regards,
                                                Vatsal

libname datalib 'D:\SAS\Datalib\';

* put pointer to input file name here;
filename in1 ('D:\Data\Data Files-2\Jul0902.txt') lrecl=2500;

filename out 'D:\SAS\Output\temp_oct301.txt';

* output graphics file in PS format ;
filename outfile 'D:\SAS\Output\cluster.ps';

        goptions device=pscolor gsfname=outfile gsfmode=append;
        options ls=90;

data datalib.tmpjul9; retain time 0;

infile in1 delimiter=','; input col1-col342;

time=col1; hrs=time/60/24;
time=time/60;
time=time/60;
time=time/24;

keep time;

file out; retain count 0;

options linesize=256;



*One Time Only;

if mod(_n_,10)=0;

put col1-col342;

proc standard mean=0 std=1 out=stan;
var col3 col9  col11 col12;
run;
proc fastclus data=stan out=clust maxclusters=5;
var col3 col9 col11 col12;
run;
proc candisc data=clust out=Can noprint;
class Cluster;
var col3 col9 col11 col12 ;

legend1 frame cframe=ligr label=none cborder=black
        position=center value=(justify=center);
        axis1 label=(angle=90 rotate=0) minor=none;
        axis2 minor=none;

        proc gplot data=Can;
                plot Can2*Can1=Cluster/frame cframe=ligr
                        legend=legend1 vaxis=axis1 haxis=axis2;
        run;
quit;



--
"Work like you don't need money, Love like you have never been hurt ,
And Dance like no one is watching. "

Contact
-------
Office : 919-401-0299 -Ext 312
Home   : 919-382-1950