topic Code does not work within a macro in SAS Programming

Code does not work within a macro

desireatem — Fri, 17 Jan 2014 19:19:36 GMT

Hello, I need the baseline covariates fro PHREG for each of the random sample with replacement. THis code does see to work well. The problem is with

baseline covariates=Pattern_&pt out=survs_&u survival= survival;.

NB: I need to run this within a macro because I have some matrices to compute. Or within a for loop.

CODE.

data neww;

seed=-1;

alpha1 = 10;

beta1 = 10;

beta2=10;

do i = 1 to 141;

Age=40+(5)*rannor(1);

lambdaT = 0.025; *baseline hazard;

lambdaC= .03; *heavy=0.15; *light=0.03;

er=0+(0.01)*rannor(1);

t = rand("WEIBULL", 0.75, lambdaT); * time of event;

c = rand("WEIBULL", 1.25, lambdaC) ;* time of censoring;

monset = min(t, c); * which came first?;

obs=(t lt c); * creating observation vaiable from censored when observeed obs=1;

pibflr= alpha1 + beta1*t + beta2*Age +er;

output;

end;

run;

%let pt=pt;

%macro pat(patt=);

%do &pt=1 %to &patt; /*** I NEED THE BASELINE COVARIATE FOR AGE=0, So for each data below new_1, new_2, I want the first to set the covariates=Pattern_1, covariates=Pattern_2*/

data Pattern_&pt;

age=0;

output;

run;

%end;

%mend pat;

%MACRO DO_BRANCH(replicates=);

%do u=1 %to &replicates;

proc surveyselect data=neww method = urs sampsize = 141 OUTHITS

reps=1 seed=1234100 out=new_&u;

run;

%pat(patt=20);

ods trace on /listing;

ods graphics on;

ods output ParameterEstimates=CPHH_&u;

proc phreg data=new_&u ;

model monset*obs(0)=age;

*baseline out=survs survival= survival;

baseline covariates=Pattern_&pt out=survs_&u survival= survival; /* The issue is that SAS doesnot seem to understand that I want covariates=pattern_1 when running the data set new_1 and covariates=pattern_2 for new_2*/

run;

ods trace off;

%end;

%MEND DO_BRANCH;

%DO_BRANCH(replicates=20);

THIS CODE RUNS BUT NOT THE ABOVE. Below is what I want the macro to do for each case.

data Pattern;

Age=0;

output;

run;

ods trace on /listing;

ods graphics on;

ods output ParameterEstimates=CPHH;

proc phreg data=new ;

model monset*obs(0)=age;

baseline covariates=pattern out=survs survival= survival ;

run;

ods trace off;

Re: Code does not work within a macro

ballardw — Fri, 17 Jan 2014 19:26:55 GMT

Are you getting errors or identical output?

If it is identical output it is likely because the seed value in the surveyselect is selecting the same records for each dataset.

Try doing a proc compare on two of the new_ datasets.

A correction would be to increment the seed value using your U macro variable.

Re: Code does not work within a macro

desireatem — Fri, 17 Jan 2014 20:12:08 GMT

I corrected the proc surveyselect, the data set are different but the code is not working;

data neww;

seed=-1;

alpha1 = 10;

beta1 = 10;

beta2=10;

do i = 1 to 141;

Age=40+(5)*rannor(1);

lambdaT = 0.025; *baseline hazard;

lambdaC= .03; *heavy=0.15; *light=0.03;

er=0+(0.01)*rannor(1);

t = rand("WEIBULL", 0.75, lambdaT); * time of event;

c = rand("WEIBULL", 1.25, lambdaC) ;* time of censoring;

monset = min(t, c); * which came first?;

obs=(t lt c); * creating observation vaiable from censored when observeed obs=1;

pibflr= alpha1 + beta1*t + beta2*Age +er;

output;

end;

run;

%let pt=pt;

%macro pat(patt=);

%do &pt=1 %to &patt; /*** I NEED THE BASELINE COVARIATE FOR AGE=0, So for each data below new_1, new_2, I want the first to set the covariates=Pattern_1, covariates=Pattern_2*/

data Pattern_&pt;

age=0;

output;

run;

%end;

%mend pat;

%MACRO DO_BRANCH(replicates=);

%do u=1 %to &replicates;

proc surveyselect data=neww method = urs sampsize = 141 OUTHITS

reps=1 samprate=1 seed=-1 out=new_&u;

run;

%pat(patt=2);

ods trace on /listing;

ods graphics on;

ods output ParameterEstimates=CPHH_&u;

proc phreg data=new_&u ;

model monset*obs(0)=age;

*baseline out=survs survival= survival;

run;

ods trace off;

%end;

%MEND DO_BRANCH;

%DO_BRANCH(replicates=2);

proc compare base=New_1 compare=New_2 printall;

title 'Comparing Two Data Sets: Full Report';

run;

data Pattern;

Age=0;

output;

run;

ods trace on /listing;

ods graphics on;

ods output ParameterEstimates=CPHH;

proc phreg data=neww;

model monset*obs(0)=age;

baseline covariates=pattern out=survs survival= survival ;

run;

ods trace off;

Re: Code does not work within a macro

Reeza — Fri, 17 Jan 2014 23:19:14 GMT

Does this do what you expect it do?

Test it %put &pt afterwards to see what it resolves to:


%let pt=pt;

This will create a set of pattern_1, pattern_2 datasets that are identical. Why? There is no point to this, simply have one pattern dataset.



%macro pat(patt=);
%do &pt=1 %to &patt; 
data Pattern_&pt;
age=0;
output;
run;
%end;
%mend pat;



%MACRO DO_BRANCH(replicates=);
%do u=1 %to &replicates;
proc surveyselect data=neww  method = urs sampsize = 141 OUTHITS
     reps=1 samprate=1  seed=-1 out=new_&u;
run;


%pat(patt=2); /*this isn't needed, see the above comments regarding why*/
ods trace on /listing;
ods graphics on;
ods output ParameterEstimates=CPHH_&u;
proc phreg data=new_&u ;
model monset*obs(0)=age;
*baseline out=survs survival= survival;
baseline covariates=Pattern_&pt out=survs_&u survival= survival;  /* The issue is that SAS doesnot seem to understand that I want covariates=pattern_1 when running the data set new_1 and covariates=pattern_2 for new_2*/
/*Does macro variable &pt exist here, and is it what you want, from the initial question?*/
run;
ods trace off;


%end;
%MEND DO_BRANCH;


%DO_BRANCH(replicates=2);

 /*Why compare your random samples?*/
proc compare base=New_1 compare=New_2 printall;
   title 'Comparing Two Data Sets: Full Report';
run;