Date:         Fri, 14 Sep 2007 17:25:34 -0500
Reply-To:     Mary <mlhoward@avalon.net>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Mary <mlhoward@AVALON.NET>
Subject:      Re: Identifying the first instance that various conditions are
              satisfied
Comments: To: Paul Miller <pmiller@OHTN.ON.CA>
Content-Type: text/plain; charset="iso-8859-1"

OK, I've done it :-).  I had fun trying this in IML this afternoon; thanks for the brain teaser!

Here's my try-

 
Mary Howard

Research Assistant III

Dept. of Ophthalmology

Univ. of Iowa Hospitals and Clinics



******

 

data set1;

informat id 2. x 1. y 1. first_x 1. first_y 1. first_xyw 1. first_xya 1. ;

input id x y;

cards;

01 1 0 

01 0 1

01 1 1 

02 1 0 

02 0 1

02 1 1 

;

 

data set1a;

   set set1;

   first_x=0;

   first_y=0;

   first_xyw=0;

   first_xya=0;

 

proc sql;

      select count(*) 

      into :set1n

      from set1a;

run;

 

%Put >>>&set1n<<< ;

 

proc iml symsize=20000 worksize=20000;

   start mod1(set1n);

   edit set1a;

   id=99999;

   x=0;

   y=0;

   count_x=0;

   count_y=0;

   count_xy=0;

   do i=1 to set1n;

      first_x=0;

        first_y=0;

        first_xyw=0;

        first_xya=0;

      old_id=id;

      read point i var 'id' into id;

        print id old_id;

        print count_x count_y count_xy;

        if id[1] ^= old_id[1] then

         do;

            count_x=0;

            count_y=0;

            count_xy=0;

          end;

      print count_x count_y count_xy;

      read point i var 'x' into x;

        if x=1 then

          do;

               if count_x = 0 then

                    first_x=1;

               count_x=count_x + 1;

            end;

        read point i var 'y' into y;

        if y=1  then

           do;

                if count_y=0 then

                   first_y=1;

                  count_y=count_y + 1;

             end;

        if x=1 & y=1 then

           do;

                if count_xy=0 then

                   first_xyw=1;

                  count_xy=count_xy+ 1;

             end;

      if (count_x=1) & (count_y >=1) then

             first_xya=1;

      else if (count_y=1) & (count_x >=1) then

             first_xya=1;

      replace; 

   end;

   finish;

   run mod1(&set1n);

quit;

run;

 

proc print data=set1a;

run;

 

  ----- Original Message ----- 
  From: Paul Miller 
  To: SAS-L@LISTSERV.UGA.EDU 
  Sent: Friday, September 14, 2007 1:25 PM
  Subject: Identifying the first instance that various conditions are satisfied


  Hello Everyone,

  I'm trying to do some coding that involves identifying the first time
  that various conditions are satisfied. Below are "have" and "need"
  datasets that represent a simplified version of what I'm trying to do.

  data have;
          input id $ x y;
          cards;
          1       1       0
          1       0       1
          1       1       1
          ;
  run;

  data need;
          input id $ x y first_x first_y first_xyw first_xya;
          cards;
          1       1       0       1       0       0       0
          1       0       1       0       1       0       1
          1       1       1       0       0       1       0
          ;
  run;

  For first_x and first_y, I just want to identify the first time that
  these variables had a value of 1 in the data. For first_xyw, I want to
  identify the first time that x and y had a value of 1 within the same
  row. For first_xya, I want to identify the first time that x and y had
  taken a value of 1 in any row. So this variable is coded 1 in row 2,
  because x = 1 in row 1 and y = 1 in row 2. Thus, it's the first time
  that we see that combination across any of the rows.

  I can find some inelegant ways of doing this (e.g., for first_xya,
  creating count variables to identify the number of times x or y has
  occurred, coding xya across rows, and then creating a count of xya so
  that I can identify the first instance).

  I was hoping to find a more elegant approach though.

  Can anyone see a good way of doing this?

  Thanks,

  Paul

  Paul J. Miller, Ph.D.
  Research Scientist and Statistician
  Ontario HIV Treatment Network
  1300 Yonge St., Suite 308
  Toronto, Ontario M4T 1X3
  Phone: (416) 642-6486 ext 232
  Fax: (416) 640-4245