topic Re: How to skip data step processing (output if the condition met) in SAS Programming

How to skip data step processing (output if the condition met)

webart999ARM — Tue, 16 Aug 2022 11:38:55 GMT

Hello everybody.

Today I faced with an issue which solution seems pretty straightforward, but I couldn't find any correct approach for it.

I have the below data

What I want to do is following.

1 - To output an additional raw (with certain new variables) if my (have) dataset dd2 does not contain observation with DDTESTCD='PRCDTH'

2 - And in case if it contains DDTESTCD='PRCDTH', then I want a new dd3 dataset with 0 observation,

Thank you in advance

Kind Regards

Artur

Re: How to skip data step processing (output if the condition met)

PaigeMiller — Tue, 16 Aug 2022 11:54:32 GMT

You can do this via a macro variable which I have named &FLAG ... something like this outline of the code

%let flag=0;
data _null_;
    set dd2;
    if ddtestcd='PRCDTH' then do; 
        call symputx('flag',1);
        stop;
    end;
run;

%if &flag=1 %then %do;
    /* Create data set with zero rows */
    data dd3;
        set dd2(obs=0);
    run;
%end;
%else %do;
    /* Some other data step or PROC code, I'm not sure what you want here */
%end;

Re: How to skip data step processing (output if the condition met)

webart999ARM — Tue, 16 Aug 2022 14:56:13 GMT

Thank you for your answer @PaigeMiller
I applied the code to my program, but I see some issues. Hope you can help me to figure it out.

%macro prcdth();
%let flagone=0; %let flagtwo=0;
data _null_;
    set dd2;
    if ddtestcd='PRCDTH' then do; 
        call symputx('flagone',1);
        stop;
    end;
run;
data _null_;
    set dd2;
    if  DDTESTCD="CANDUSO" and DDORRES="UNEQUIVOCALLY DUE TO DISEASE UNDER STUDY" then do; 
        call symputx('flagtwo',1);
        stop;
    end;
run;

%if &flagone.=1 %then %do;    /* Create data set with zero rows */
    data dd3_1;
        set dd2(obs=0);
    run;
%end;
%else %if &flagtwo.=1 %then %do;
    data dd3_2;
	set dd2;
	 
 	AVALC = "CANCER";	
	
   aval=.;
   PARAMCD="PRCDTH";
   AVALCAT1='NON-CARDIOVASCULAR CAUSE';
        if AVALCAT1='CARDIOVASCULAR CAUSE'     then AVALCA1N=1;
   else if AVALCAT1='NON-CARDIOVASCULAR CAUSE' then AVALCA1N=2;
   output;
run;
%end;
%else %if &flagone.^=1 and &flagtwo.^=1 %then %do;
data dd3_3;
	set dd2;

	if DTHFL='Y'  then do;
	
	AVALC="Unknown";
	 aval=.;
   PARAMCD="PRCDTH";
   AVALCAT1=DDSCAT;
        if AVALCAT1='CARDIOVASCULAR CAUSE'     then AVALCA1N=1;
   else if AVALCAT1='NON-CARDIOVASCULAR CAUSE' then AVALCA1N=2;
output;
end;
run;
%end;
%mend prcdth;
%prcdth();

 1          OPTIONS NONOTES NOSTIMER NOSOURCE NOSYNTAXCHECK;
 61         
 62         
 63         %macro prcdth();
 64         %let flagone=0; %let flagtwo=0;
 65         data _null_;
 66             set dd2;
 67             if ddtestcd='PRCDTH' then do;
 68                 call symputx('flagone',1);
 69                 stop;
 70             end;
 71         run;
 72         data _null_;
 73             set dd2;
 74             if  DDTESTCD="CANDUSO" and DDORRES="UNEQUIVOCALLY DUE TO DISEASE UNDER STUDY" then do;
 75                 call symputx('flagtwo',1);
 76                 stop;
 77             end;
 78         run;
 79         
 80         %if &flagone.=1 %then %do;    /* Create data set with zero rows */
 81             data dd3_1;
 82                 set dd2(obs=0);
 83             run;
 84         %end;
 85         %else %if &flagtwo.=1 %then %do;
 86             data dd3_2;
 87         set dd2;
 88         
 89          AVALC = "CANCER";
 90         
 91            aval=.;
 92            PARAMCD="PRCDTH";
 93            AVALCAT1='NON-CARDIOVASCULAR CAUSE';
 94                 if AVALCAT1='CARDIOVASCULAR CAUSE'     then AVALCA1N=1;
 95            else if AVALCAT1='NON-CARDIOVASCULAR CAUSE' then AVALCA1N=2;
 96            output;
 97         run;
 98         %end;
 99         %else %if &flagone.^=1 and &flagtwo.^=1 %then %do;
 100        data dd3_3;
 101        set dd2;
 102        
 103        if DTHFL='Y'  then do;
 104        
 105        AVALC="Unknown";
 106         aval=.;
 107           PARAMCD="PRCDTH";
 108           AVALCAT1=DDSCAT;
 109                if AVALCAT1='CARDIOVASCULAR CAUSE'     then AVALCA1N=1;
 110           else if AVALCAT1='NON-CARDIOVASCULAR CAUSE' then AVALCA1N=2;
 111        output;
 112        end;
 113        run;
 114        %end;
 115        %mend prcdth;
 116        %prcdth();
 MPRINT(PRCDTH):   data _null_;
 MPRINT(PRCDTH):   set dd2;
 MPRINT(PRCDTH):   if ddtestcd='PRCDTH' then do;
 MPRINT(PRCDTH):   call symputx('flagone',1);
 MPRINT(PRCDTH):   stop;
 MPRINT(PRCDTH):   end;
 MPRINT(PRCDTH):   run;
 
 NOTE: There were 15 observations read from the data set WORK.DD2.
 NOTE: DATA statement used (Total process time):
       real time           0.01 seconds
       cpu time            0.01 seconds
       
 
 MPRINT(PRCDTH):   data _null_;
 MPRINT(PRCDTH):   set dd2;
 MPRINT(PRCDTH):   if DDTESTCD="CANDUSO" and DDORRES="UNEQUIVOCALLY DUE TO DISEASE UNDER STUDY" then do;
 MPRINT(PRCDTH):   call symputx('flagtwo',1);
 MPRINT(PRCDTH):   stop;
 MPRINT(PRCDTH):   end;
 MPRINT(PRCDTH):   run;
 
 NOTE: There were 42 observations read from the data set WORK.DD2.
 NOTE: DATA statement used (Total process time):
       real time           0.00 seconds
       cpu time            0.01 seconds
       
 
 MPRINT(PRCDTH):   data dd3_1;
 MPRINT(PRCDTH):   set dd2(obs=0);
 MPRINT(PRCDTH):   run;
 
 NOTE: There were 0 observations read from the data set WORK.DD2.
 NOTE: The data set WORK.DD3_1 has 0 observations and 47 variables.
 NOTE: DATA statement used (Total process time):
       real time           0.00 seconds
       cpu time            0.00 seconds
       
 
 117        
 118        OPTIONS NONOTES NOSTIMER NOSOURCE NOSYNTAXCHECK;
 131

So I have added two additional derivations and &flagtwo macro variable.

The second derivation (dd3_2 dataset) I need to execute when the &flagtwo is 1 and after the &flagone else condition (i.e there is no ddtestcd='PRCDTH' in dd2 dataset, but there is an observation satisfying DDTESTCD="CANDUSO" and DDORRES="UNEQUIVOCALLY DUE TO DISEASE UNDER STUDY" condition).
The third derivation (dd3_3 dataset) I need to execute when the &flagone is ^=1 and the &flagtwo is ^=1, after the &flagtwo else condition (i.e. there is no ddtestcd='PRCDTH' in dd2 dataset and also there is no observation satisfying DDTESTCD="CANDUSO" and DDORRES="UNEQUIVOCALLY DUE TO DISEASE UNDER STUDY" condition, but there is an observation satisfying DTHFL='Y' condition).

Before %prcdth macro execution the input dataset dd2 has 1170 observations(checked manually).
But during the macro execution it takes only 15 and 42 observations, and dd3_2 and dd3_3 datasets are not created.

Could you please take a look and give some advice on what need to be adjusted, I have attached the log as well.

Thank you in advance.

Re: How to skip data step processing (output if the condition met)

PaigeMiller — Tue, 16 Aug 2022 15:13:40 GMT

But during the macro execution it takes only 15 and 42 observations

The STOP command makes the DATA step stop processing observations, so the first time it stops at 15 and the next time it stops at 42. This is what I expect.

and dd3_2 and dd3_3 datasets are not created.

That's because you wrote

%else %if &flagtwo.=1 %then %do;

Take the %else out of this line. And remove %else later as well.

Re: How to skip data step processing (output if the condition met)

webart999ARM — Tue, 16 Aug 2022 16:38:59 GMT

Take the %else out of this line. And remove %else later as well

All these three conditions - %if &flagone.=1, %if &flagtwo.=1 and %if &flagone.^=1 and &flagtwo.^=1 are mutually exclusive but dependent from each other. If I remove the %else, then they will be executed independently from each other correct?

second(&flagtwo) condition should be executed when the &flagone. condition is false, and the third condition should be executed when the first two conditions are not met.

So at least(and only) one of them should be executed.

Re: How to skip data step processing (output if the condition met)

PaigeMiller — Tue, 16 Aug 2022 21:14:14 GMT

All these three conditions - %if &flagone.=1, %if &flagtwo.=1 and %if &flagone.^=1 and &flagtwo.^=1 are mutually exclusive but dependent from each other. If I remove the %else, then they will be executed independently from each other correct?

Please try it and find out for yourself.

second(&flagtwo) condition should be executed when the &flagone. condition is false, and the third condition should be executed when the first two conditions are not met. So at least(and only) one of them should be executed.

Somewhere, there is a contradiction between your explanation above, and your statement that you want data sets dd3_2 and dd3_3 output. So I don't understand what you want and cannot help until I understand. But in any case, this is a simple matter of coming up with the proper %IF %THEN %ELSE or %IF %THEN %IF %THEN (without %ELSE). Surely, since you know what the desired result is, you can state it clearly in %IF %THEN %ELSE or %IF %THEN %IF %THEN (without %ELSE) language.

Re: How to skip data step processing (output if the condition met)

webart999ARM — Wed, 17 Aug 2022 06:40:29 GMT

Thank you