How do i output in a macro a unique output file/dataset for each cycle?

[Dcicantab5]

Hi,

How do i output in a macro a unique output file/dataset for each cycle, to the temporary work folder, so that it will be available for use in a second macro?

%macro speed(univariatecate, factor);
	proc logistic data=work.sd;
		class gender &univariatecate. (ref="0")/param=ref;
		model Outcome(event='1')=Age_yr_ 
			gender &univariatecate. &factor./selection=none;
		output out=sdcrossvalidm2 predprobs=crossvalidate;
	run;

%mend;

%speed(bleed, PR);
%speed(bleed, ALT);
%speed(bleed, AST);

%macro speedcv(univariatecate, factor);
	proc logistic data=work.sdcrossvalidm2 plots(only label)=all;
		class gender &univariatecate. (ref="0")/param=ref;
		model Outcome(event='1')=Age_yr_ 
			gender &univariatecate. &factor./selection=none;
		roc pred=xp_1;
		roccontrast;
		ods output rocassociation=SDcrossvalidatedm2;
	run;

%mend;

%speed(bleed, PR);
%speed(bleed, ALT);
%speed(bleed, AST);
ods graphics off;

In the code above (which don't work), the first macro do not output unique files for each cycle {(bleed,PR), (bleed,ALT), (bleed,AST)}. I want to make unique files so that it will be available to the second macro which crossvalidates the first.

Help & guidance most appreciated!

Thank you,

Saiful.

[Reeza]

Give your output files unique names....come up with some sort of naming convention that you can use in your second macro. 

OR transpose your data so that PR/ALT/AST are by groups and then use BY group processing. 

https://communities.sas.com/t5/SAS-Communities-Library/How-do-I-write-a-macro-to-run-multiple-regres...

[Patrick · Accepted solution]

As for the suggestion to create unique table names, code as below used at the begining of your first macro could do the job.

  %global my_dsname;
  %let my_dsname=t_%sysfunc(datetime(),b8601dt15.0);

And then in your 2nd macro once you're done with the data set and to keep WORK tidy, you could delete the ds using code as below:

proc datasets lib=%scan(work.&my_dsname,-2,.) nolist nowarn;
  delete %scan(&my_dsname,-1,.);
run;quit;

[Dcicantab5]

Thanks guys, both advices got me thinking, hence the following worked!:

%macro speed(univariatecate, factor, my_dsname);
	proc logistic data=work.sd;
		class gender &univariatecate. (ref="0")/param=ref;
		model Outcome(event='1')=Age_yr_ 
			gender &univariatecate. &factor./selection=none;
		output out=&my_dsname. predprobs=crossvalidate;
	run;

%mend;

%speed(bleed, PR, bleedpr);
%speed(bleed, ALT, bleedalt);
%speed(bleed, AST, bleedast);


%macro speedcv(univariatecate, factor, my_dsname);
	proc logistic data=&my_dsname. plots(only label)=all;
		class gender &univariatecate. (ref="0")/param=ref;
		model Outcome(event='1')=Age_yr_ 
			gender &univariatecate. &factor./selection=none;
		roc pred=xp_1;
		roccontrast;
	run;

%mend;

%speedcv(bleed, PR, bleedpr);
%speedcv(bleed, ALT, bleedalt);
%speedcv(bleed, AST, bleedast);