Make a loop to calculate bootstrap 95% CI for median

[mariko5797]

I got the following code from online and was wondering if there was a way to efficiently run it multiple times with different combinations of PARAMN, APHASE, and DAYN. I assume a do loop will get me there but I am not at all familiar with how to set one up.

%let NumSamples = 1000;

data sample;
 set _trt_;
 where paramn = 1 and aphase = 'CHMI 1' and dayn = 1;
 keep aval;

proc surveyselect data= sample noprint seed= 1
 out= _boot_ (rename= (replicate = sampleid))
 method= urs		/*resample with replacement*/
 samprate= 1 		/*each bootstrap sample has N observations*/
 reps= &NumSamples;	/*generate # bootstrap resamples*/

proc means data= _boot_ noprint;
 by sampleid;
 freq NumberHits;
 var aval;
 output out= _boot_med_ median= med;

proc univariate data= _boot_med_ noprint;
 var med;
 output out= _95ci_ pctlpre= CI95_ 
 		pctlpts= 2.5 97.5
		pctlname= lb ub;

data _111;
 set _95ci_;
 	statn= 5;
	ci = strip(lb)||', '||strip(ub);
 	paramn = 1; aphase = 'CHMI 1'; dayn = 1;
run;

Mock Dataset:

data have0;
 input usubjid $ paramn aphase $ dayn aval @@;
 cards;
 A	1	CHMI1	1	0.45
 A	1	CHMI1	2	0.36
 A	1	CHMI1	3	0.66
 A	2	CHMI1	1	0.25
 A	2	CHMI1	2	0.58
 A	2	CHMI1	3	0.55
 A	2	CHMI1	4	0.65
 A	3	CHMI1	1	0.25
 A	3	CHMI1	2	0.85
 A	1	CHMI2	1	0.12
 A	1	CHMI2	2	0.54
 A	1	CHMI2	3	0.98
 A	2	CHMI2	1	0.13
 A	3	CHMI2	1	0.21
 A	3	CHMI2	2	0.48
 ;
run;

data have;
 set have0; output;
 set have0; usubjid = 'B'; aval = aval + 0.05; output;
 set have0; usubjid = 'C'; aval = aval - 0.1; output;
run;

[mariko5797]

EDIT: the sample dataset will come from HAVE not _TRT_ in this case

[PaigeMiller · Accepted solution]

You could do this yourself, or you could use the existing %BOOT and %BOOTCI macros, and save yourself a lot of programming time.

https://blogs.sas.com/content/iml/2018/07/23/boot-and-bootci-macros-sas.html

[mariko5797]

Although that would help me reduce the bootstrap code itself down, I would still need to create different combinations of PARAMN, APHASE, and DAYN to get all the 95% CI that I want. This is the main reason I would like to make a loop.