Dynamically create variables in the data step based on values of another variable

[Zelazny7]

I regularly have to process flags from comma-separated character fields that look like this:

example_field = "AX,EQ,XX,TY,M,GY,TS,Z,WP,DL";

My current process involves the following steps:

1. Data step that iterates through the comma-separated values and outputs each one into a new dataset
2. Proc freq of those values and store them into macro variables
3. A second data step that checks for the presence of the values and assigns them to a binary variable flag

I am wondering whether all of this can be reduced to a single datastep.  I have dozens of variable like this and I would like to create a macro that can dynamically create flags for each comma-separated value in one datastep.  Something like this:

data tmp;   

     example_field = "AX,EQ,XX,TY,M,GY,TS,Z,WP,DL";   

     %expand_list(example_field);

/* yields the output:

     flag_AX

     flag_EQ

     flag_XX

     ...

     flag_DL

*/

run;

Is this wishful thinking?  I understand the timing issues between the data step and the macro compilation steps so I think I may be out of luck.  I'm hoping someone can give me a glimmer of hope, though.  Maybe something like proc FCMP?  The flag doesn't have to be strictly [1,0], it could also be [1, .]

[Reeza]

Post sample data and sample output I think to help understand the problem.

I think you could at least combine step 2/3 using a proc transpose step instead of macro variables, but would need to see more info to be sure.

[scottdmurff]

Have you looked at the SYMPUT and SYMGET routines to see if they can help you accomplish what you are after.  See here for documentation: SAS(R) 9.2 Macro Language: Reference

[Zelazny7]

Yes, thanks, I've looked into those.  The problem I'm having is that the variables I want to create derive their names from the values of another field.  SYMGET will assign a macro variable value to a variable I have already created, but it won't create the variable for me dynamically.

[Reeza · Accepted solution]

Bad example data so hard to determine if this is correct or not:

data have;

example_field = "AX,EQ,XX,TY,M,GY,TS,Z,WP,DL";

output;

run;

data want1;

    set have;

    i=1;

    do while (scan(example_field, i, ",") ne "");

    var_flip=scan(example_field, i, ",");

    value=1;

    output;

    i+1;

    end;

run;

proc transpose data=want1 out=want2 prefix=flag_;

id var_flip;

var value;

run;