Date:         Fri, 20 Feb 2004 15:34:09 -0500
Reply-To:     diskin.dennis@KENDLE.COM
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Dennis Diskin <diskin.dennis@KENDLE.COM>
Subject:      Re: Help needed with outputting..
Comments: To: Nagakumar Sridhar <nsridhar@ATHEROGENICS.COM>
Content-Type: text/plain; charset="us-ascii"

Kumar,

To automate something like this you pretty much have to go to a macro. The
macro can retrieve the actual treatments into macro variable and then use
them to build a datastep. For this to work, the treatment names must be
valid as dataset names. E.g. not start with numbers or contain spaces or
other invalid characters.

A sample macro: (I don't have time to test right now)

%macro split(data=);
proc sql noprint;
select distinct trim(trtmnt) into :ds1 thru :ds999 from &data where trtmnt
ne ' ';
%let n_ds = &sqlobs;
quit;

data nulls
%do i = 1 %to &n_ds;
   &&ds&i
%end;
;
set &data;
select(trtmnt);
  %do i = 1 %to &n_ds;
     when("&&ds&i") output &&ds&i;
   %end;
  otherwise output nulls;
end;
run;
%mend;

%split(data=yourdataset);

HTH,
Dennis Diskin






Nagakumar Sridhar <nsridhar@ATHEROGENICS.COM>
Sent by: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
02/20/2004 03:21 PM
Please respond to Nagakumar Sridhar


        To:     SAS-L@LISTSERV.UGA.EDU
        cc:
        Subject:        Help needed with outputting..


Hi Folks:



Here's the (new!) question.  I have a dataset like so:



Obs.
Analyte                     trtmnt             visit              Status
N



1453    WBC (THOUS/M)    ds20mg
DAY 5       I       1

1454    WBC (THOUS/M)    ds20mg
DAY 5       NC      4

1455    WBC (THOUS/M)    ds320mg
DAY 5       NC      5

1456    WBC (THOUS/M)    ds40mg
DAY 5       NC      6

1457    WBC (THOUS/M)    ds5mg
DAY 5       NC      6

1458    WBC (THOUS/M)    ds640mg
DAY 5       NC      6

1459    WBC (THOUS/M)    ds80mg
DAY 5       NC      6

1460    WBC (THOUS/M)    dsplacebo
DAY 5       NC      5

1461    WBC (THOUS/M)    ds10mg
DAY 6       NC      5

1462    WBC (THOUS/M)    ds160mg
DAY 6       NC      6

1463    WBC (THOUS/M)    ds20mg
DAY 6       D       1

1464    WBC (THOUS/M)    ds20mg
DAY 6       I       1

1465    WBC (THOUS/M)    ds20mg
DAY 6       NC      4

1466    WBC (THOUS/M)    ds320mg
DAY 6       NC      5

1467    WBC (THOUS/M)    ds40mg
DAY 6       NC      6

1468    WBC (THOUS/M)    ds5mg
DAY 6       NC      6

1469    WBC (THOUS/M)    ds640mg
DAY 6       NC      6

1470    WBC (THOUS/M)    ds80mg
DAY 6       NC      6

1471    WBC (THOUS/M)    dsplacebo
DAY 6       NC      5

1472    WBC (THOUS/M)    ds10mg
DAY 7       NC      5

1473    WBC (THOUS/M)    ds160mg
DAY 7       NC      6

1474    WBC (THOUS/M)    ds20mg
DAY 7       D       1

1475    WBC (THOUS/M)    ds20mg
DAY 7       I       2

1476    WBC (THOUS/M)    ds20mg
DAY 7       NC      3

1477    WBC (THOUS/M)    ds320mg
DAY 7       NC      5

1478    WBC (THOUS/M)    ds40mg
DAY 7       NC      6

1479    WBC (THOUS/M)    ds5mg
DAY 7       NC      6

1480    WBC (THOUS/M)    ds640mg
DAY 7       NC      6

1481    WBC (THOUS/M)    ds80mg
DAY 7       NC      6

1482    WBC (THOUS/M)    dsplacebo
DAY 7       I       1

1483    WBC (THOUS/M)    dsplacebo
DAY 7       NC      4





What I want to do is to output  all the ds20mg to a dataset ds20mg and
all ds40mg's to ds40mg and so on.  If the treatment

Groups were defined (that is, if I knew how many treatment groups there
were or what they were called) life would be easy.  But that's not the
case.  The number is dynamic and so's the treatment groups.



Question is how would I be able to output the various observations to
the different datasets, given the above conditions?



Thanks and regards



Kumar