Summing all numeric variables (flexible names and amount of variables)

[Quadraaa]

Wow, gentlemen, I am really surprised with how detailed and useful your answers are   When I see these tricks I have two contradictory feelings: 1) I feel upset about my poor knowledge of SAS and 2) I feel glad that there are so many interesting things to learn

Thanks a lot to all of you.

[FriedEgg]

I have to send a kudos to SASJedi as well for going well beyond in his response what I would consider necessary and giving you not just an answer to point you in a direction but a real process to utilize.

[slsnx2]

I have to share my code:

I know it's a closed topic but I think this could be useful.

/************************************************************************************/

/* Descripcion general del programa:                                                    */

/* Suma las variables numericas de una tabla, sin especificar nombres. */

/* Autor             : Saul Sanchez Sánchez                                             */

/************************************************************************************/

data a;

do i=1 to 100;

  a=abs(int(rand('NORMAL')*1000));

  b=abs(int(rand('NORMAL')*1000));

  c=abs(int(rand('NORMAL')*1000));

  cl='a'||substr(left(compress(i)),1,1);

  output;

end;

run;

data vars(where=(type='N'));

   length name $ 8 type $ 1;

   drop dsid i num rc;

   dsid=open("work.a","i");

   num=attrn(dsid,"nvars");

   do i=1 to num;

      name=varname(dsid,i);

      type=vartype(dsid,i);

      output;

   end;

   rc=close(dsid);

run;

data vars(keep=function);

set vars;

function='sum('||compress(name)||') as '||compress(name);

run;

proc sql;

select function into :function separated by ','

from vars;

quit;

%put &function;

proc sql;

create table b as

select cl, &function

from a

group by 1;

run;