topic Re: Grouping observations using one by variable in SAS Programming

Grouping observations using one by variable

GN0001 — Wed, 22 Sep 2021 16:48:03 GMT

Hello team,

Why a grouping is not happening if I don't bring first and last in grouping? What is the role of first and last?

Proc sort data=cert.usa out=work.temp;
by dept;
run;
data work.budget (Keep = dept payroll);
set work.temp;
by dept;
if wagecate = 'S' then yearly = wagerate*12;
else if wagecat='H" then yearly = wagerate*2000;
if first.dept then payroll=0;
payroll+yearly;
if last.dept;
Run;

Re: Grouping observations using one by variable

PeterClemmensen — Wed, 22 Sep 2021 16:49:56 GMT

What do you mean by "grouping is not happening" ?

Re: Grouping observations using one by variable

PaigeMiller — Wed, 22 Sep 2021 17:15:11 GMT

@GN0001 wrote:

Hello team,

Why a grouping is not happening if I don't bring first and last in grouping? What is the role of first and last?

First and last create the grouping. If you don't have it, then no grouping is created because there is no code to do that grouping.

FIRST.DEPT sets the sum to zero each time DEPT changes. LAST.DEPT is used to control which records are output, only the last record for each value of DEPT is output.

I always recommend that you don't do summing (or computing most other statistics) by group in a DATA step, because SAS has already written that code for you in PROC MEANS/PROC SUMMARY. Really, don't do this in a DATA step.

proc summary data=temp nway;
    class dept;
    var payroll;
    output out=sums sum=;
run;

Re: Grouping observations using one by variable

Kurt_Bremser — Wed, 22 Sep 2021 18:20:22 GMT

Maxim 1: Read the Documentation.

See How SAS Identifies the Beginning and End of a BY Group and BY-Group Processing in the DATA Step