Date:         Thu, 28 Aug 2008 11:32:43 -0400
Reply-To:     Jonathan Goldberg <jgoldberg@BIOMEDSYS.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Jonathan Goldberg <jgoldberg@BIOMEDSYS.COM>
Subject:      Re: How to Create a flag for this logic in single datastep

Al:

It's not entirely clear what you want.  However, if it's one flag record
per subj, then this:

data need;
 retain hasdos 0;
 merge x
          y;
 by subj;
 if totdos > 0 then hasdos = 1;
 if last.subj then do;
  if hasdos = 1 & ^missing(tr) then flag = 'Y';
  else                              flag = 'N';
  hasdos = 0;
  output;
 end;
run;

gives this as output:

  subj    flag

   1      Y
   2      N
   3      N
   4      Y

The tool needed was the last.subj variable, which is generated (along with
the first.subj variable) automatically by SAS when a data set is read with
a "BY", as it is in the merge.  The first. and last. variables are set to
1 when the associated condition is true, to 0 otherwise.

Check out the write-up in the documentation.  It's a basic data step
programming tool, useful in many situations.

Jonathan

On Thu, 28 Aug 2008 08:07:23 -0700, Al <ali6058@GMAIL.COM> wrote:

>HI ,
>   I have two datsets x , y  .. I am supposed to merge them and
>generate the flag  based on this
>condition  if x.totdos>0 for anyday(doseno) and y.tr ne '' then Y;
>else N
>
>I am able to get this right but in multiple datasteps .. How can it be
>done in single datastep by merging
>
>
>
>Dataset X
>subj doseno totdos
>1     dose1  5
>1     dose2  5
>1     dose3  5
>2     dose1  7
>2     dose2  8
>2     dose3  9
>3     dose1  .
>4     dose1  7
>4     dose2  8
>4     dose3  9
>
>Dataset Y
>subj tr
>1    SD
>2
>4    PD
>
>
>Thanks in Advance