Date:         Tue, 18 Mar 2003 17:04:38 -0500
Reply-To:     Wenge Guo <wenge.guo@NDSU.NODAK.EDU>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Wenge Guo <wenge.guo@NDSU.NODAK.EDU>
Subject:      Re: Deleting rows (sorted columnwise)

On Mon, 17 Mar 2003 15:17:25 -0800, Paul Choate <pchoate@DDS.CA.GOV> wrote:

>It was pointed out to me that my method won't work if you have
>duplicate values, such as 2-2-2 and 3-1-1.  If this is the case, and
>the numbers are small, such as fewer than 10 each of integers 0-9, you
>could work around duplicates by using powers of 10 rather than 2.  For
>example 3011200012 would be the set 0,0,1,5,5,6,7,9,9,9, or 10^0 +
>10^0 + 10^1 + 10^5 + 10^5 + 10^6 + 10^7 + 10^9 + 10^9 + 10^9
>
>-Paul
>
>pchoate@dds.ca.gov (Paul Choate) wrote in message
news:<e6e2c481.0303171151.184b3881@posting.google.com>...
>> Richard -
>>
>> In a similar situation I've used sums of powers of 2 (2^n) where n are
>> the data values.  Note that the powers of two are 1,2,4,8,16,etc and
>> each unique subset will sum to a unique integer.  This is the same as
>> assigning a unique binary number to each set of values: your "3 1 4"
>> will be 2^3+2^1+2^4=26 decimal which is 11010 binary.  Put a counter
>> on your data, sort, and unduplicate on the first by-group observation.
>>
>> hope that helps
>>
>> Paul Choate
>>
>> >
>> > -----Original Message-----
>> > From: SAS(r) Discussion [mailto:SAS-L@listserv.uga.edu] On Behalf Of
>> > Richard A. DeVenezia
>> > Sent: Friday, March 14, 2003 2:38 AM
>> > To: SAS-L@listserv.uga.edu
>> > Subject: Q: Deleting rows (sorted columnwise)
>> >
>> > I have some data in columns A, B, C and want to delete all but first
>> > occurring row where the values are the same if the values were sorted.
>> >
>> > A B C
>> > 3 1 4
>> > 1 2 3
>> > 4 1 3 * remove (same as 3 1 4)
>> > 3 2 5
>> > 3 2 1 * remove (same as 1 2 3)
>> > 1 3 4 * remove (same as 3 1 4)
>> >
>> > Is there a way to remove the rows with out needing to sort each row by
>> > A,B,C
>> > ?





Try to run the following program. Orderal function seems to provide a
feasible solution.

data temp;
 input A B C;
 j=_n_;
 array order[*] order1-order3;
 do i=1 to 3 by 1;
  order[i]=ordinal(i,a,b,c);
 end;
 drop i;
 datalines;
 3 1 4
 1 2 3
 4 1 3
 3 2 5
 3 2 1
 1 3 4
 ;
run;

data result;
 set temp;
 by order1 order2 order3 notsorted;
 if (first.order1 and first.order2 and first.order3);
 keep a b c;
run;


best,
wenge
wenge.guo@ndsu.nodak.edu