Date:         Thu, 19 Apr 2007 04:30:18 -0700
Reply-To:     sbarry@SBBWORKS.COM
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Scott Barry <sbarry@SBBWORKS.COM>
Organization: http://groups.google.com
Subject:      Re: Dups and date - urgent
Comments: To: sas-l@uga.edu
In-Reply-To:  <1176935983.204528.43330@e65g2000hsc.googlegroups.com>
Content-Type: text/plain; charset="iso-8859-1"

On Apr 18, 6:39 pm, sdlenter <sdlent...@aol.com> wrote:
> On Apr 18, 2:55 pm, Huang...@PRINCIPAL.COM ("Huang, JS") wrote:
>
> > This is an approach:
>
> > data Given;
> >   input F1 $ F2 $  F3 $  F4 $ date:yymmdd8.;
> >   format date mmddyy10.;
> >   datalines;
> > ab   ff  gg     hg   20070505
> > ab   ff   gg    jj    20040701
> > ab  fh   ht    kkk   20040404
> > ab  ff    gg   gggg  20040508
> > ;
>
> > proc sort data=Given;
> >   by F1 F2 F3 date;
> > run;
>
> > data Result;
> >   set Given;
> >     by F1 F2 F3;
> >   if first.F3 then output;
> > run;
>
> > proc print;
> > run;
>
> > ***** Output *****
> >                                          The SAS System       16:51
> > Wednesday, April 18, 2007   4
>
> >                            Obs    F1    F2    F3     F4           date
>
> >                             1     ab    ff    gg    gggg    05/08/2004
> >                             2     ab    fh    ht    kkk     04/04/2004
>
> > -----Original Message-----
> > From: SAS(r) Discussion [mailto:S...@LISTSERV.UGA.EDU] On Behalf Of
>
> > sdlenter
> > Sent: Wednesday, April 18, 2007 4:42 PM
> > To: S...@LISTSERV.UGA.EDU
> > Subject: Dups and date - urgent
>
> > Urgent! Need help...
> > Using Dataset and Not proc SQL 'having min' function
>
> > I have sas dataset
> > F1   F2  F3   F4    date
> > ab   ff  gg     hg   20070505
> > ab   ff   gg    jj    20040701
> > ab  fh   ht    kkk   20040404
> > ab  ff    gg   gggg  20040508
>
> > I need to have a table with unique records , based on F1, F2, and F3
> > field ONLY and keeping one reccord with the earliest date among the dups
>
> > so the new table will be
> > F1   F2  F3   F4    date
> > ab  fh   ht    kkk   20040404
> > ab  ff    gg   gggg  20040508
>
> > thank you!
>
> > - Show quoted text -
>
> I cannot sort it, the dataset is too big

options compress=yes;
...your data step code to read input...
* order obs with earliest date first. ;
proc sort data=Given;
  by F1 F2 F3 date;
run;
* remove duplicates based on BY var list only. ;
proc sort nodupkey equals data=Given;
  by F1 F2 F3 ;
run;


Scott Barry
SBBWorks, Inc.