topic PROC SORT by multiple variables in New SAS User

PROC SORT by multiple variables

Rohit_R — Thu, 18 Oct 2018 09:04:30 GMT

Hi,

I have a dataset that looks like follows: I want to get the latest start for the ID, i.e. if there are 2 starts in a day for an ID, I would like to keep the later start on that day.

DATA have:

ID Startdate Startdatetime

1 1/1/2000 1/1/2000 14:12:35

1 1/1/2000 1/1/2000 14:14:35

1 1/1/2000 1/1/2000 12:12:35

1 5/1/2000 5/1/2000 14:12:35

2 1/1/2000 1/1/2000 14:12:35

2 2/1/2000 2/1/2000 14:12:35

2 2/1/2000 2/1/2000 19:12:35

3 1/1/2000 1/1/2000 14:12:35

3 2/1/2000 2/1/2000 15:12:35

3 3/1/2000 3/1/2000 16:12:35

DATA Want:

ID Startdate Startdatetime

1 1/1/2000 1/1/2000 14:14:35

1 5/1/2000 5/1/2000 14:12:35

2 1/1/2000 1/1/2000 14:12:35

2 2/1/2000 2/1/2000 19:12:35

3 1/1/2000 1/1/2000 14:12:35

3 2/1/2000 2/1/2000 15:12:35

3 3/1/2000 3/1/2000 16:12:35

Here is what I tried:

PROC SORT DATA=have;

BY ID DESCENDING StartdateTime Startdate;

RUN;

DATA want;

SET have;

BY Startdate;

IF first.Startdate;

RUN;

The sort runs fine, however, I get an error "BY variable are not properly sorted."

Any ideas on getting it right? I am not sure where I am going wrong.

Thanks,

Re: PROC SORT by multiple variables

RW9 — Thu, 18 Oct 2018 09:15:07 GMT

As you haven't put test data in the form of a datastep, this is untested. Also, please avoid coding in uppercase, its shouting at me.

proc sort data=have out=want;
  by id startdate descending startdatetime;
run;
data want;
  set want;
  by id startdate descending startdatetime;
  if first.startdate;
run;

Re: PROC SORT by multiple variables

Kurt_Bremser — Thu, 18 Oct 2018 12:57:28 GMT

The by in the sort has to exactly match the by in the data step. You may add additional variables in the sort, but up to what's used in the data step they have to be identical.