topic Re: Replicate number in PROC SURVEYSELECT in a macro in SAS Programming

Replicate number in PROC SURVEYSELECT in a macro

GKati — Tue, 10 Apr 2018 10:05:49 GMT

Dear SAS users,

I am generating samples using SURVEYSELECT and then running PROC GLM with a BY command, all of this in a MACRO to save the results in indivudal files.

I want the repetition variable, which goes from 1... n in each iteration to start where the other iteration left off.

So:

if i=1 then rep=1, ...,5

if i=2 then rep=6,...10

if i=3 then rep=11...15

etc.

I would make the REP option in my code equal to REP=(&i*&rep)+&rep which would give me this result, but that doesn't seem to be recognized as a symbol.

Any ideas on how to achieve the same result?

%macro do_survey;
options mprint symbolgen;
%do i=0 %to 2;

proc surveyselect data=claims out=bootsample
     seed = 1347 method = urs
	 samprate = 1 outhits rep =(&i*&rep)+&rep;
run;

ods listing close;

proc glm data=Bootsample;
class year(ref='2010' ) age_gr (ref='0' );
model indavgcost= year|avgcontract/ solution;
TITLE "ALL HOSPITALS-indavgcost";
by replicate;
ods output ParameterEstimates=work.t_&i;
run;

data t_&i;
set t_&i;
keep replicate parameter estimate;
run;

data t_&i;
set t_&i;
if estimate=0 then delete;
run;

proc datasets;
append base=t1 data=t_&i force;
run;
quit;

%end;
%mend do_survey;

%do_survey;

Re: Replicate number in PROC SURVEYSELECT in a macro

FloT — Tue, 10 Apr 2018 13:15:45 GMT

Hi,
What if you replace

rep =(&i*&rep)+&rep

with

reps =%eval((&i*&rep)+&rep)

... assuming that &rep is initialized somewhere ?

Re: Replicate number in PROC SURVEYSELECT in a macro

GKati — Tue, 10 Apr 2018 13:49:29 GMT

Thanks for the responses. It does run with rep=%eval()..., but it does't give me the desired result after all.

I ended up creating a new variable in a subsequent data step. This works well.

%macro do_survey;
options mprint symbolgen;
%do i=0 %to 2;

proc surveyselect data=claims out=bootsample
     seed = 1347 method = urs
	 samprate = 1 outhits rep =&rep;
run;

ods listing close;

proc glm data=Bootsample;
class year(ref='2010' ) age_gr (ref='0' );
model indavgcost= year|avgcontract/ solution;
TITLE "ALL HOSPITALS-indavgcost";
by replicate;
ods output ParameterEstimates=work.t_&i;
run;

data t_&i;
set t_&i;
keep replicate parameter estimate;
run;

data t_&i;
set t_&i;
if estimate=0 then delete;
rep2=(&i*&rep)+replicate;
run;

proc datasets;
append base=t1 data=t_&i force;
run;
quit;

%end;
%mend do_survey;

%do_survey;