Date:         Fri, 2 Apr 2004 07:47:07 -0500
Reply-To:     Nathaniel_Wooding@DOM.COM
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Nat Wooding <Nathaniel_Wooding@DOM.COM>
Subject:      Re: tricky data extraction
Comments: To: neumann <geekaxl@21CN.COM>
Content-type: text/plain; charset=us-ascii

Jim's reply was close but needed some tweaking. In particular, not that one
needs a "notsorted" on the by statement.

Nat Wooding

data a;
   input obs var1 var2;

   cards;
1 1 1.1001113E17
2 2 20031231
3 1 1.1001113E17
4 2 20040110
5 2 20040108
6 2 20040108
7 2 20040108
8 1 1.1001113E17
9 2 20040110
10 1 1.1001113E17
11 2 20040110
12 1 1.1001113E17
13 2 20040110
14 2 20031213
15 2 20031222
16 2 20031222
17 2 20031225
data b;
    set a;
      by var1 notsorted;
    if (not first.var1 and var1 =2) then delete;

proc print;run;



                      neumann
                      <geekaxl@21CN.COM        To:       SAS-L@LISTSERV.UGA.EDU
                      >                        cc:
                      Sent by: "SAS(r)         Subject:  tricky data extraction
                      Discussion"
                      <SAS-L@LISTSERV.U
                      GA.EDU>


                      04/02/04 01:18 AM
                      Please respond to
                      neumann






Obs VAR1 VAR2
1 1 1.1001113E17
2 2 20031231
3 1 1.1001113E17
4 2 20040110
5 2 20040108
6 2 20040108
7 2 20040108
8 1 1.1001113E17
9 2 20040110
10 1 1.1001113E17
11 2 20040110
12 1 1.1001113E17
13 2 20040110
14 2 20031213
15 2 20031222
16 2 20031222
17 2 20031225

hi all,

      i have a dataset listed upside, what i want to do is to extract
every observation pair whose VAR1 is '1' and '2', when the VAR1 equals
1, it's a ID number , and when VAR1 equals 2, it's a meaningful record
which i want to use, the rest should be deleted,and when it's done, it
should be like this:


Obs VAR1 VAR2
1 1 1.1001113E17
2 2 20031231
3 1 1.1001113E17
4 2 20040110
8 1 1.1001113E17
9 2 20040110
10 1 1.1001113E17
11 2 20040110
12 1 1.1001113E17
13 2 20040110

anyone knows how to get this done? i need your suggestions, Thanx in
advance!