topic Re: Covert char to numeric in SAS Programming

Covert char to numeric

rajd1 — Fri, 01 Oct 2021 15:41:01 GMT

Hi all,

I have a character variable that needs to be converted to numeric. The character <0.001 needs to be converted to 0.001 numeric value.

Data Have:

data have;
input iD nums $;
cards;
1	1.675	
2	<0.001	
3	3.578	
4	<0.00001
;

Re: Covert char to numeric

Reeza — Fri, 01 Oct 2021 15:42:41 GMT

Would you also have cases of >= or <= or other symbols that need to be accounted for in this situation?

@rajd1 wrote:

Hi all,

I have a character variable that needs to be converted to numeric. The character <0.001 needs to be converted to 0.001 numeric value.

Data Have:
data have;
input iD nums $;
cards;
1	1.675	
2	<0.001	
3	3.578	
4	<0.00001
;

Re: Covert char to numeric

rajd1 — Fri, 01 Oct 2021 15:44:41 GMT

@Reeza :

Yes, there might be < or > <= or >=

Re: Covert char to numeric

Reeza — Fri, 01 Oct 2021 15:45:48 GMT

Always at the beginning?

Re: Covert char to numeric

Patrick — Fri, 01 Oct 2021 16:02:32 GMT

What will work depends on your data. If you're sure that there can't be numbers in scientific notation then the derivation syntax for nums_new_1 should be suitable.

data have;
  input iD nums $;
  cards;
1 1.675 
2 <0.001  
3 3.578 
4 <0.00001
;

data want;
  set have;
  nums_new_1=input(compress(nums,'.+-','kd'),best32.);
  nums_new_2=input(compress(nums,'<=> '),best32.);
run;

Re: Covert char to numeric

ballardw — Fri, 01 Oct 2021 15:57:29 GMT

@Reeza wrote:

Would you also have cases of >= or <= or other symbols that need to be accounted for in this situation?
@rajd1 wrote:

Hi all,

I have a character variable that needs to be converted to numeric. The character <0.001 needs to be converted to 0.001 numeric value.

Data Have:
data have;
input iD nums $;
cards;
1	1.675	
2	<0.001	
3	3.578	
4	<0.00001
;

One way might be:

data want;
  set have;
  newnum=input(compress(nums, "+-.", "kd"),12.);
run;

The compress function removes everything except the + - decimal point and digits (the d option indicates digits, the k keeps instead of removing. Then input converts to numeric. This assumes you don't have more than 12 characters in the resulting numeric value string.

Warning: if for some reason you have exponential notation values such as 1.5E3 in the data the E gets removed and the result would be a missing value. If that is the case you need to check for the presence of the E and treat that string differently (add E to the list of characters "+-.eE" for example)