Date:         Fri, 30 Aug 2002 08:50:13 -0400
Reply-To:     Asheber Sewalem <sewalem@CDN.CA>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
Comments:     To: "Nevseta, Floyd G" <Floyd.G.Nevseta@bankofamerica.com>,
              "Dorfman, Paul" <Paul.Dorfman@BCBSFL.COM>
From:         Asheber Sewalem <sewalem@CDN.CA>
Subject:      Re: transpose
Comments: To: SAS-L@LISTSERV.VT.EDU
Content-Type: text/plain; charset="iso-8859-1"

Hi Floyd,Paul,Prasad and all,


Thanks for your reply,

   proc freq data=a;
   tables code * score / nocol norow nopercent;
   run;

This is fine this sample data set. However, if the data set is huge with
many variables, I usually get an error message. That is why I was trying
to avoid using proc freq.

In addition to that I need the actual output file. Therefore, Paul's
suggestion is taken as it does it in one data step.

Many thanks
Asheber






This is the easiest way:

proc freq data=a;
tables code * score / nocol norow nopercent;
run;

Regards,
Floyd

-----Original Message-----
From: Asheber Sewalem [mailto:sewalem@CDN.CA]
Sent: Thursday, August 29, 2002 3:33 PM
To: SAS-L@LISTSERV.UGA.EDU
Subject: transpose


Hi all,

I would appreciate if you give  me an efficient way of doing the
following data set.(i.e., the total number of observation in each code
and by score). I used proc transpose and it did not give me the desired
output, may be I missed something. Please have a look at below the
desired output.


data a;
input id code score;
cards
;
161    12    3
139    12    2
120    12    1
163    12    2
144    10    3
143    10    3
135    10    2
132    11    2
147    11    2
143    13    3
170    13    1
155    14    1
;

The desired output should be,

Obs    code  score_1 score_2 score_3   total
 1      10      0     1         2       3
 2      11      0     2         0       2
 3      12      1     2         1       4
 4      13      1     0         1       2
 5      14      1     0         0       1

Asheber