topic Re: converting character to numeric variable in SAS Programming

converting character to numeric variable

Senthil1102 — Thu, 11 Oct 2012 05:49:10 GMT

i had a character variable with the following values...
1.0
2.123
1.222
1.0000
3.1022

while converting it to numeric variable, the trailing zeros are disappeared...is there any way to retain the exact decimal numbers while converting from character to numeric variable...

2.123

1.222

3.1022

Re: converting character to numeric variable

Tom — Thu, 11 Oct 2012 11:22:18 GMT

SAS stores numbers as floating point, so there is no difference in how it is stored between the numbers 1.0 and 1.0000 in your example input stream. You can use a FORMAT to tell SAS how to display the number, but the same format will be applied to every observation in the data. For your example data you could use the format 6.4 (which can also be written as F6.4).

data test;

input char $ @@;

num = input(char,6.4);

put char= num= num= 6.4 ;

cards;

1.0 2.123 1.222 1.0000 3.1022

run;

char=1.0 num=1 num=1.0000

char=2.123 num=2.123 num=2.1230

char=1.222 num=1.222 num=1.2220

char=1.0000 num=1 num=1.0000

char=3.1022 num=3.1022 num=3.1022

Re: converting character to numeric variable

data_null__ — Thu, 11 Oct 2012 12:53:37 GMT

You can count the number of decimals in the character string. What you do with that information is up to you.

data dec;
   input c $;
   if index(c,'.') 
      then d = length(scan(c,-1,'.'));
      else d = 0; 
   num = input(c,f12.);
   cards; 
1.0
2.123
1.222
1.0000
3.1022
;;;;
   run;

Re: converting character to numeric variable

Senthil1102 — Thu, 11 Oct 2012 14:01:49 GMT

is any other way to retain the decimai numbers specifically 1.0 and 1.0000 during the conversion from character to numeric?

Re: converting character to numeric variable

jbaldwin — Thu, 11 Oct 2012 21:19:51 GMT

To follow-on what data_null_ stated ("What you do with that information is up to you.") I think it would be helpful to describe why you want to do this. It would seem that if you need it both numeric and character, then you keep both variables and use the one that's needed at the time.

For example if the objective is to line up the decimal places so that you get something like

1.000

31.2

0.00012220000

Then you would need to create an additional character string with a fixed length and the decimal would always show in the same position of the character string.