Date:         Wed, 24 Sep 2003 11:46:55 -0400
Reply-To:     Robert Abelson <rabelson@KAI-RESEARCH.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Robert Abelson <rabelson@KAI-RESEARCH.COM>
Subject:      Re: proc transpose?
Content-Type: text/plain

        This is easily done in a data step:

        data new(keep=id date dx);
            set old;
            array _dx[*] dx1-dx3;
            do i=1 to hbound(_dx);
                dx = _dx[i];
                if dx ne " " then output;
            end;
        run;

Bob Abelson
KAI Research, Inc.
6001 Montrose Rd.
Suite 920
Rockville, MD 20852
T: 301-770-2730
F: 301-770-4183
rabelson@kai-research.com

>                       jcern
>                       <jcern15@YAHOO.C         To:
> SAS-L@LISTSERV.UGA.EDU
>                       OM>                      cc:
>                       Sent by: "SAS(r)         Subject: proc transpose?
>                       Discussion"
>                       <SAS-L@LISTSERV.
>                       UGA.EDU>
>
>
>                       09/24/2003 09:46
>                       AM
>                       Please respond
>                       to jcern
>
>
>
>
>
>
> I am trying to transpose some variables into observations and for some
> reason proc transpose isn't really working.  My data looks something
> like this:
>
> ID  Date     DX1      Dx2      Dx3
> 001 1/1/00   789.0    453.1    678.4
> 001 1/1/00   675.0    V57.0
> 001 3/24/01  453.2    234.9
>
> I want the output to look like this:
>
> ID  DATE    DX
> 001 1/1/00 789.0
> 001 1/1/00 453.1
> 001 1/1/00 678.4
> 001 1/1/00 675.0
> 001 1/1/00 V57.0
> 001 3/24/01 453.2
> 001 3/24/01 234.9
>
> For some reason the proc transpose is transposing BOTH ways in the
> same step: it is creating VARIABLES as well as new observations.  Not
> sure why...
>
> Thanks!