topic Handling zero sample size in proc surveryselect in SAS Programming

Handling zero sample size in proc surveryselect

KJazem — Tue, 14 Mar 2023 13:16:21 GMT

I have a macro that takes a random sample of a dataset at every run and outputs a table. For example, if the random sample size is 1000, and if the source table somehow has fewer than 1000 records, it takes all the records.

%let _sampsize = <code that chooses a sample size based on rules>
proc surveyselect data=CAD.SOURCE out=CAD.RAND_DEST
    method=srs sampsize=&_sampsize. selectall;
run;

But if the _samplesize macro variable is for whatever reason, zero, the code breaks and says sampsize has to be a positive integer. What would be the best/most reliable way to proof this?

I would prefer to return an empty table if the sample size is zero. I guess I check against sample size first, but is there a more 'elegant' way?

Re: Handling zero sample size in proc surveryselect

sbxkoenk — Tue, 14 Mar 2023 13:16:21 GMT

Just enhance your macro logic such that the PROC SURVEYSELECT with a zero sample size will never be executed.
You know how to do this?

Koen

Re: Handling zero sample size in proc surveryselect

KJazem — Tue, 14 Mar 2023 13:19:03 GMT

Yes, this is my first thought. But I would like to return an empty table instead.

if &_sampsize. = 0 then do;
data CAD.RAND_DEST;
set CAD.SOURCE;
run;
end;

Would this work? Each output table (random sample) will be appended to main table.

Re: Handling zero sample size in proc surveryselect

sbxkoenk — Tue, 14 Mar 2023 13:31:47 GMT

This would work :

if &_sampsize. = 0 then do;
 data CAD.RAND_DEST;
  set CAD.SOURCE;
  IF 0; /* subsetting IF , but never true */
 run;
end;

Koen

Re: Handling zero sample size in proc surveryselect

Tom — Tue, 14 Mar 2023 14:32:45 GMT

@KJazem wrote:
Yes, this is my first thought. But I would like to return an empty table instead.

if &_sampsize. = 0 then do;
data CAD.RAND_DEST;
set CAD.SOURCE;
run;
end;

Would this work? Each output table (random sample) will be appended to main table.

If the goal is to APPEND observations to a table and you want zero observations appended, then just skip the APPEND step.

If you want to generate a dataset with zero observations you could add a STOP statement.

data CAD.RAND_DEST;
  stop;
  set CAD.SOURCE;
run;

Or the OBS=0 dataset option.

data CAD.RAND_DEST;
  set CAD.SOURCE (obs=0);
run;