topic Re: Call Execute in SAS Programming

Call Execute

DoumbiaS — Tue, 11 Jul 2017 12:59:35 GMT

Hello,

One can not invoke a macro using Call execute in oder to handle a created global macro var by call sypmut.

How is that ? Any minds is welcome.

Regards

Re: Call Execute

PaigeMiller — Tue, 11 Jul 2017 13:13:35 GMT

I think the documentation explains how to do that

http://documentation.sas.com/?cdcId=pgmmvacdc&cdcVersion=9.4&docsetId=mcrolref&docsetTarget=n1q1527d51eivsn1ob5hnz0yd1hx.htm&locale=en

Re: Call Execute

Kurt_Bremser — Tue, 11 Jul 2017 13:20:26 GMT

Post your code.

This works:

%macro mymac(varname);
%put Variable as received by macro: &&&varname;
%mend;

data _null_;
call symputx('myvar','mytext','g');
call execute('%mymac(myvar)');
run;

Log from that:

24         %macro mymac(varname);
25         %put Variable as received by macro: &&&varname;
26         %mend;
27         
28         data _null_;
29         call symputx('myvar','mytext','g');
30         call execute('%mymac(myvar)');
31         run;

Variable as received by macro: mytext
NOTE: DATA statement used (Total process time):
      real time           0.00 seconds
      cpu time            0.00 seconds

Re: Call Execute

DoumbiaS — Tue, 11 Jul 2017 14:43:14 GMT

Look at this example: val_m2 values are not updated form call execute.

Regards

data have;
do i=1 to 10;
	have= i+1;
	output;
end;
run;

%macro mymacro(nc);
%Global used mg1 mg2;
%let mg1= 20; %let mg2= 40;
data need; set have nobs= nobs; J= _n_;
var0= J * &mg1 / &mg2;     				
var1= var0 + 0.5 * &mg1 / &mg2;
var2= var0 + var1;
nexp= var2 * &mg1 / &mg2;
need + (((nexp - have)**2) / nexp);           /*cumulative sum*/ 
if J= nobs then	call symput('used',need); 
run;
%mend;

data see;
do i= 1 to 3;
	if i= 1 then do;
		max= 40;  min= 15;  m1= 27.5; 
		val_m1= 2;  m2= 33.75;   
		call execute('%mymacro('||m2||')');
		val_m2= input(symget('used'), 30.);    /*updating*/              
	end;
	else do;
		min= m1;  m1= m2;  val_m1= val_m2;  m2= (m2 + max)/2;   /*updating*/		
		call execute('%mymacro('||m2||')');
		val_m2 = input(symget('used'), 30.);
	end;
output;
end;
run;

Re: Call Execute

Kurt_Bremser — Tue, 11 Jul 2017 15:10:41 GMT

The data step in the macro can only be compiled and run after the data step with the call execute has finished. Macro code in a macro that is call executed, OTOH, will be resolved immediately when pushed into the SAS interpreter queue.

If you want to use data from a second dataset dynamically, but in a way that can not be handled by a join, consider using a hash object.

Since you only use an overall statistic from dataset have, I'd prepare that beforehand. No need to reread the whole dataset in every iteration.

Re: Call Execute

DoumbiaS — Tue, 11 Jul 2017 15:19:54 GMT

Thanks,

An overall statistics is not used. When the macro execute, it use the current variable have value.

Re: Call Execute

DoumbiaS — Tue, 11 Jul 2017 15:21:46 GMT

Also, the macro executes with the last m2 value as argument.

Re: Call Execute

Kurt_Bremser — Tue, 11 Jul 2017 15:24:37 GMT

Since dataset 'have' never changes, and only the results of the final iteration of the data step are put into the macro variable, and the code in the macro is not dynamic at all, there is no need to call the macro at all. Just set &need once.

If you plan on doing something completely different, show THAT code.

Re: Call Execute

ballardw — Tue, 11 Jul 2017 15:50:17 GMT

Where you use these two lines in the data step:

call execute('%mymacro('||m2||')'); <= creates a call to the macro that executes AFTER the data step
val_m2= input(symget('used'), 30.); <= so the value of USED is not changed prior to this line. It will be the same value for every iteration

Do you intentionally through away and decimal part of used?

Re: Call Execute

DoumbiaS — Wed, 12 Jul 2017 08:58:47 GMT

No i don't thinnk so.

Your explanations are so clear, thanks for that.