topic Re: Divide population into macro variables in SAS Programming

Divide population into macro variables

Kiteulf — Thu, 09 Sep 2021 11:34:19 GMT

I want to divide up a population into cohorts of subpopulations where I have the numbers in a macro var.



data Population;
do ids = 1 to 100000; 
output;
end;
;
run;

proc sql noprint;
            select ceil(count(*)/7500)
            into :cohort
            from Population;
      quit;
%put &cohort;

%macro test;
%do i = 1 %to &cohort;
  
            proc sql noprint;
                  select distinct ids
                  into :KSIds 
                  from Population(firstobs = &i cohort = &i);
            quit;
 %end;
%mend;

First I am finding out how many cohorts I need, then I am trying to get a macro to write no more than 7500 KSids from the Population file into the KSIds0, KSIds1. KSids2 etc

I just

Re: Divide population into macro variables

PaigeMiller — Thu, 09 Sep 2021 11:47:34 GMT

Macros not needed, this can be done in a data step

data cohorts;
    set population;
    cohort = floor((ids-1)/7500);
run;

Re: Divide population into macro variables

Patrick — Thu, 09 Sep 2021 12:02:47 GMT

Or as a variant to what @PaigeMiller proposes.

data Population;
  do ids = 1 to 100000; 
    output;
  end;
  stop;
run;

data cohorts;
  set population;
  cohort = mod(_n_,7500)+1;
run;

Re: Divide population into macro variables

Kiteulf — Thu, 09 Sep 2021 12:06:06 GMT

I think I am misunderstood, (my poor explanation to blame)

I want the ids to be in a list

each macro containing ids

first cohort : id 1-7500

second cohort: id 7501-15000

Re: Divide population into macro variables

PaigeMiller — Thu, 09 Sep 2021 12:28:54 GMT

@Kiteulf wrote:

I think I am misunderstood, (my poor explanation to blame)

I want the ids to be in a list

each macro containing ids

first cohort : id 1-7500

second cohort: id 7501-15000

I don't know why you need them in a separate list, or actually many separate lists, when they are clearly identified in the data set I named COHORT.

What would you do with these macro variable lists once you have it? Please explain the next step. I'm relatively sure that the data set COHORT can provide what you need. Whatever the next step is, I don't think you need macro variables containing long lists of IDs.