topic Re: Detecting letters in variables in SAS Programming

Detecting letters in variables

NilsNoh — Wed, 25 Jul 2018 09:01:42 GMT

Hey Guys,

this question was probably already asked before but I am not only new to SAS, but also new to this forum, so please bear with me.

I have a dataset where I wanted to sort a character variable with length 18, although it is a character variable it only contains numbers (IDs). If I try to sort this variable, I get an error because in at least one observation there are letters mixed into the ID.

How do I find out how many observations contain letters, I think I have to use the anyalpha command, but not quite sure how to apply it.

Thanks in advance!

Re: Detecting letters in variables

RW9 — Wed, 25 Jul 2018 09:15:45 GMT

Can you use sortseq=linguistic (numeric_collation=on) in your proc sort?

Anyhows:

number_of_chars=lengthn(compress(id," ","ka"));

Pop that in a datastep and it will tell you how many alphas are in the id string. ka = keep alpha.

Re: Detecting letters in variables

Amir — Wed, 25 Jul 2018 10:57:59 GMT

Hi,

To find the number of observations, you could try proc sql as in the following example:

/* set up an id variable with some containing alphabetic characters */
data have;
   set sashelp.class;

   length id $ 8;

   /* add alphabetics to id if name begins with 'J' */
   if name eq: 'J' then
      id = cats('ABC',id);
   else
      id = cats(_n_);
run;

%let id_alpha = 0;

/* save the count of id's with alphabetic characters into a macro variable */
proc sql noprint;
   select
      count (*)
   into
      :id_alpha
   from
      have(keep = id)
   where
      anyalpha(id)
   ;
quit;

%put id_alpha = &id_alpha;

Regards,

Amir.

Re: Detecting letters in variables

MadhuKorni — Wed, 25 Jul 2018 11:12:16 GMT

Anyalpha is used to find the first occurrence of an alphabet in a string.

data want;
set have;
retain cnt 0;;
Num_of_Chars = lengthn(compress(ID,,'d')); *If you want to include special characters(if exist);
Num_of_Chars = lengthn(compress(ID,,'ka')); *If you have only alphabets and digits;
if Num_of_Chars gt 0 then cnt=cnt+1;
call symput("No_Of_Obs",cnt);
run;

Re: Detecting letters in variables

NilsNoh — Wed, 25 Jul 2018 11:40:44 GMT

Cheers Guys!

Tried the methods and both work well and deliver the same result, thank you!

Re: Detecting letters in variables

Amir — Wed, 25 Jul 2018 12:05:25 GMT

Thanks @NilsNoh.

You can mark either solution as accepted.

Regards,

Amir.