DATA Step, Macro, Functions and more

how to pick variable names

Accepted Solution Solved
Reply
Contributor
Posts: 36
Accepted Solution

how to pick variable names

Dear SAS community:

I have a large dataset with several thousand variables. 

I need to pick variables using specific texts in their variable names. 

I use this simple code to pick them:

 

data two; set one;

keep mi:;

run;

 

this picks up the variables that starts with the name "mi", such as mi_date, mi_size, and mi_treat, but cannot pick up hx_mi, time_mi, and chronic_mi_treat.  is there something simple I can use to pick out variables using "mi" in the variable names regardless of position?

 

thanks

 

 


Accepted Solutions
Solution
‎05-06-2017 09:07 AM
Super User
Posts: 5,082

Re: how to pick variable names

Something simple?  No.  But the complexities are manageable.  Old-style program:

 

proc contents data=have out=_contents_ (keep=name where=(index(upcase(name), 'MI') > 0));

run;

 

That gives you a list of all the variable names.  Note that you can use dictionary.columns instead of PROC CONTENTS if that's something y ou are mildly familiar with.

 

Then get the list into a macro variable (again you have choices at this point, CALL EXECUTE being one alternative):

 

proc sql;

select strip(name) into : varlist separated by ' ' from _contents_;

quit;

 

Finally, use the macro variable:

 

data want;

set have (keep=&varlist);

run;

 

Note that you can add to the list of variables, such as:

 

set have (keep=id &varlist);

 

Finally, consider whether it would be safer (but still accurate) to search for "mi_" instead of "mi".

View solution in original post


All Replies
Solution
‎05-06-2017 09:07 AM
Super User
Posts: 5,082

Re: how to pick variable names

Something simple?  No.  But the complexities are manageable.  Old-style program:

 

proc contents data=have out=_contents_ (keep=name where=(index(upcase(name), 'MI') > 0));

run;

 

That gives you a list of all the variable names.  Note that you can use dictionary.columns instead of PROC CONTENTS if that's something y ou are mildly familiar with.

 

Then get the list into a macro variable (again you have choices at this point, CALL EXECUTE being one alternative):

 

proc sql;

select strip(name) into : varlist separated by ' ' from _contents_;

quit;

 

Finally, use the macro variable:

 

data want;

set have (keep=&varlist);

run;

 

Note that you can add to the list of variables, such as:

 

set have (keep=id &varlist);

 

Finally, consider whether it would be safer (but still accurate) to search for "mi_" instead of "mi".

Contributor
Posts: 36

Re: how to pick variable names

many thanks Astounding,

fantastic, this is simple enough

thanks

 

☑ This topic is SOLVED.

Need further help from the community? Please ask a new question.

Discussion stats
  • 2 replies
  • 211 views
  • 1 like
  • 2 in conversation