Date:         Wed, 18 Sep 2002 16:16:22 -0400
Reply-To:     Howard_Schreier@ITA.DOC.GOV
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Howard_Schreier@ITA.DOC.GOV
Subject:      Re: selecting records
Content-Type: text/plain; charset=ISO-8859-1

But suppose there were a seventh record with the same Vn values as the
third record:

  7   A  B  A  10

As I interpret the problem, it too should be deleted. So I would use the
following code for the third step:

   data w2;
   set w;
   by v1 v3 v4;
   if first.v4 then v2match=v2;
   if               v2match=v2;
   retain           v2match;
   drop             v2match;
   run ;

On Wed, 18 Sep 2002 14:49:08 -0400, Ian Whitlock <WHITLOI1@WESTAT.COM>
wrote:

>Sebastian,
>
>Here is code based on diskin.dennis@KENDLE.COM suggestion to sort the file.
>
>data w ; input
>rec v1 $ v2 $ v3 $ v4 $ ; cards ;
>  1   A  A  A  10
>  2   A  A  A  10
>  3   A  B  A  10
>  4   B  A  B  20
>  5   B  B  B  40
>  6   C  A  B  50
>;
>
>proc sort data = w ;
>   by v1 v3 v4 rec ;
>run ;
>
>data w2 ;
>   set w ;
>   lv1 = lag ( v1 ) ;
>   lv2 = lag ( v2 ) ;
>   lv3 = lag ( v3 ) ;
>   lv4 = lag ( v4 ) ;
>
>   if v1=lv1 and v3=lv3 and v4=lv4 then
>   do ;
>      if v2 ^= lv2 then delete ;
>   end ;
>run ;
>
>proc sort data = w2 ;
>   by rec ;
>run ;
>
>IanWhitlock@westat.com
>
>-----Original Message-----
>From: Sebastien Dube [mailto:Sebastien.Dube@MCGILL.CA]
>Sent: Wednesday, September 18, 2002 1:57 PM
>To: SAS-L@LISTSERV.UGA.EDU
>Subject: Re: selecting records
>
>
>Sorry for the confusion. You were right, somthing was wrong in my
>question. Let me rephrase that.
>
>My data would look like this (I switched the symbols...)
>
> rec# v1 v2 v3 v4
>  1   A  A  A  10
>  2   A  A  A  10
>  3   A  B  A  10
>  4   B  A  B  20
>  5   B  B  B  40
>  6   C  A  B  50
>
>looking specifically at the first 3 records, I wish to discard #3
>because for v1, v3 and v4 it has the same values as record #1 AND a
>different value in V2 (than record #1). However, I wish to keep record
>#2 because the value in v2 is the same as for record #1. In sum, record
>#3 is the only one I wish to delete for this dataset.
>
>Thank you again.
>Sébastien
>
>
>
>--
>Sébastien Dubé
>Planning Analyst
>
>University Planning Office
>James Administration Bldg. rm: 633
>McGill University
>845 Sherbrooke W St.
>Montreal, Quebec
>H3A 2T5
>
>Phone:  (514)398-7072
>Fax:    (514)398-7358