Date:         Mon, 21 Aug 2006 07:50:24 +0200
Reply-To:     Gadi Ben-Avi <gadi_b@MALAM.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Gadi Ben-Avi <gadi_b@MALAM.COM>
Organization: Bezeq International Ltd.
Subject:      Re: URGENT: PROC SORT TO PROC SQL CONVERSION
Comments: To: sas-l@uga.edu

Chris,

At a certain point an options was added to proc sort that lets you get the
'deduped' observations in a separate dataset.

The DUPOUT option will create a dataset with the deduped observations.

Gadi
"ChrisW75" <9Squirrels@gmail.com> wrote in message
news:1156118064.027338.32280@75g2000cwc.googlegroups.com...
> Best practice for deduping is to not use either Proc SQL or Proc Sort I
> reckon.  This is my method:-
>
> proc sort data=dset1;
>  by A B C D E;
> run;
>
> data dset2
>        dupes;
>  set dset1;
>  by A B C D E;
>
>  if first.E then output dset2;
>  if not(first.E and Last.E) then output dupes;
> run;
>
> This gives you a way to look at your duplicate records after the
> dedupe.
> ChrisW75
>
> Scott Bass wrote:
>> I agree with Ken, see this thread in the archive (read the entire
>> threaed):
>> http://www.listserv.uga.edu/cgi-bin/wa?A2=ind0605A&L=sas-l&D=0&P=1162
>>
>> <kenneth_m_lin@sbcglobal.net> wrote in message
>> news:rm1Gg.9471$%j7.1256@newssvr29.news.prodigy.net...
>> > Unless your dataset only contains the fields A through E, it'd be
>> > difficult to perform de-duping in PROC SQL.
>> >
>> > "Abhijat" <ursabhijat@gmail.com> wrote in message
>> > news:1156068234.842470.228350@i42g2000cwa.googlegroups.com...
>> >> Hi All,
>> >> Please someone suggest me a proc sql equivalent for the below
>> >> mentioned
>> >> proc sort code.
>> >>
>> >>
>> >> PROC SORT DATA = abc.xyz OUT  = abc.pqr NODUPKEY;
>> >> BY A B C D E;
>> >> RUN;
>> >>
>> >>
>> >> Many Thanks ,
>> >> Abhijat.
>> >>
>> >
>> >
>