James,

Using Ian's own old trick, it can be done even more economically:

data c (drop=f); do until (last.studid); set a; by studid; if f then continue; output; if grade = 70 then f = 1; end; run;

It will run just a bit more efficiently, because 1) the flag is reset at the end of each by-group automatically (which is just a tad faster than resetting it explicitly) 2) the bottom and top of the step are reached the number of times equal to the number of by-groups instead of number of observations, hence reducing the amount of housekeeping. Plus, there is no need to retain anything :-).

Kind regards, =================== Paul M. Dorfman Jacksonville, Fl ===================

>From: James Yang <jamesy99@HOTMAIL.COM> >Reply-To: James Yang <jamesy99@HOTMAIL.COM> >To: SAS-L@LISTSERV.UGA.EDU >Subject: Re: Help needed! >Date: Mon, 10 Jul 2000 14:14:40 EDT > >Thank you all very much: Ian, Ya, Nancy, Nat, Rob, Puddin', Kelly. I >learned >a lot from you experts. Thanks again! > >James > > >>From: Ian Whitlock <WHITLOI1@WESTAT.com> >>To: 'Ya Huang' <ya.huang@AGOURON.COM> >>CC: 'James Yang' <jamesy99@HOTMAIL.COM> >>Subject: RE: Help needed! >>Date: Mon, 10 Jul 2000 13:43:10 -0400 >> >>Ya, >> >>You should do it with one flag, then the code is easier to understand. >> >>data xx; >> set xx; >> retain outflg 1; >> by studid; >> if first.studid then outflg = 1 ; >> if outflg then output ; >> if grade =70 then outflg = 0 ; >>run ; >> >>Ian Whitlock <whitloi1@westat.com> >> >>-----Original Message----- >>From: Ya Huang [mailto:ya.huang@AGOURON.COM] >>Sent: Monday, July 10, 2000 1:17 PM >>To: SAS-L@LISTSERV.UGA.EDU >>Subject: Re: Help needed! >> >> >>James, try the following: >> >>data xx; >>input studID $ GRADE; >>cards; >>01 90 >>01 96 >>01 70 >>01 50 >>01 70 >>02 80 >>02 70 >>02 90 >>02 76 >>; >> >>data xx; >> set xx; >>retain flag1 flag2; >>by studid; >>if first.studid then do; flag1=0; flag2=0; end; >>if grade =70 then do; flag1=1; flag2+1; end; >>if flag2 gt 0 then flag2+1; >>if flag1 and flag2 > 2 then delete; >> >>options nocenter; >> >>proc print; >>run; >> >>--------------------------------------- >>OBS STUDID GRADE FLAG1 FLAG2 >> >> 1 01 90 0 0 >> 2 01 96 0 0 >> 3 01 70 1 2 >> 4 02 80 0 0 >> 5 02 70 1 2 >> >> >> >>James Yang wrote: >> > >> > I have following dataset: J1 with 2 vars: >> > >> > StudID GRADE >> > 01 90 >> > 01 96 >> > 01 70 >> > 01 50 >> > 01 70 >> > 02 80 >> > 02 70 >> > 02 90 >> > 02 76 >> > >> > What I want to do now is: for a given student, all records after the >>first >> > time grade=70 are deleted. That is, for this specific example, >>following >> > will be my expected output dataset: >> > >> > StudID GRADE >> > 01 90 >> > 01 96 >> > 01 70 >> > 02 80 >> > 02 70 >> > >> > How can make this efficiently? Any advice is appreciated in advance. >> > >> > James

________________________________________________________________________ Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com