3361 Data test; 3362 var1=8; var2=0; do i=1 to 1E5; output; end; 3363 var2=1; do i=1 to 1E5; output; end; 3364 3365 var1=9; var2=0; do i=1 to 1E5; output; end; 3366 var2=1; do i=1 to 1E5; output; end; 3367 3368 var1=10; var2=0; do i=1 to 1E5; output; end; 3369 var2=1; do i=1 to 1E5; output; end; 3370 Run;

NOTE: The data set WORK.TEST has 600000 observations and 3 variables. NOTE: DATA statement used (Total process time): real time 0.59 seconds cpu time 0.56 seconds

3371 3372 proc sql _method ; 3373 create table maxTest0 as 3374 select max(var1) as maxVar1 3375 from test 3376 where var2=0 3377 ;

NOTE: SQL execution methods chosen are:

sqxcrta sqxsumn sqxsrc( WORK.TEST ) NOTE: Table WORK.MAXTEST0 created, with 1 rows and 1 columns.

3378 quit; NOTE: PROCEDURE SQL used (Total process time): real time 1.18 seconds cpu time 0.29 seconds

-----Original Message----- From: Keintz, H. Mark [mailto:mkeintz@wharton.upenn.edu] Sent: Thursday, May 01, 2008 3:28 PM To: Sigurd Hermansen Subject: RE: Re: Logic help

Sigurd:

To be a complete role model of efficiency, you might want to do this in the DATA TEST step:

Data test; var1=8; var2=0; do i=1 to 1E5; output; end; var2=1; do i=1 to 1E5; output; end;

var1=9; var2=0; do i=1 to 1E5; output; end; var2=1; do i=1 to 1E5; output; end;

var1=10; var2=0; do i=1 to 1E5; output; end; var2=1; do i=1 to 1E5; output; end; Run;

-----Original Message----- From: SAS(r) Discussion [mailto:SAS-L@LISTSERV.UGA.EDU] On Behalf Of Sigurd Hermansen Sent: Thursday, May 01, 2008 1:14 PM To: SAS-L@LISTSERV.UGA.EDU Subject: Re: Logic help

Another method that does not appear to require sorting or specialized filtering works no matter where the maximum appears in a dataset: data test; do i=1 to 1E5; var1=8; var2=0; output; end; do i=1 to 1E5; var1=8; var2=1; output; end; do i=1 to 1E5; var1=9; var2=0; output; end; do i=1 to 1E5; var1=9; var2=1; output; end; do i=1 to 1E5; var1=10; var2=0; output; end; do i=1 to 1E5; var1=10; var2=1; output; end; run; proc sql _method ; create table maxTest0 as select max(var1) as maxVar1 from test where var2=0 ; quit;

Relational database methods in particular define solutions that do not depend on physical ordering of data. For this problem a logical solution, as opposed to a procedural one, doesn't suffer much (if any) from performance inefficiencies:

NOTE: SQL execution methods chosen are: sqxcrta sqxsumn sqxsrc( WORK.TEST ) NOTE: Table WORK.MAXTEST0 created, with 1 rows and 1 columns.

3360 quit; NOTE: PROCEDURE SQL used (Total process time): real time 1.28 seconds cpu time 0.34 seconds S

-----Original Message----- From: owner-sas-l@listserv.uga.edu [mailto:owner-sas-l@listserv.uga.edu] On Behalf Of Ya Huang Sent: Wednesday, April 30, 2008 7:24 PM To: SAS-L@LISTSERV.UGA.EDU; Deep Singh Cc: Ya Huang Subject: Re: Logic help

You are very close:

data want; set have (where=(var2=0)) end=last; if last then output; run;

On Wed, 30 Apr 2008 19:15:23 -0400, DP <adsingh78@GMAIL.COM> wrote:

>Hi, > >How to get the maximum event time (MAXEVETM) that should be the last >value of VAR1 where VAR2=0. > >Have: >VAR1 VAR2 >8 0 >8 0 >8 1 >8 0 >8 1 >9 0 >9 0 >9 0 >9 1 >10 0 >11 1 > >Want: > >VAR1 VAR2 MAXEVETM >10 0 10 > >Can it be done without sorting the data? > >I have this but not sure if correct… > >data want; > set have end=last; > if last and CENSOR=0 then output; >run; > >Thanks a lot! > >DP