Date:         Sat, 22 Jun 1996 15:45:17 EDT
Reply-To:     John Whittington <100517.3677@COMPUSERVE.COM>
Sender:       "SAS(r) Discussion" <SAS-L@UGA.CC.UGA.EDU>
From:         John Whittington <100517.3677@COMPUSERVE.COM>
Subject:      Re: reading character variable as numeric
Comments: To: Ramu Thiagarajan <ramut@NWU.EDU>

On Sat, 22 Jun 1996, Ramu Thiagarajan <ramut@NWU.EDU> wrote:
 
>        I have a data set wherein one of the variables has a character
>        suffix which is redundant.  Example: 0 is coded as 0_ (i.e.,
>        there is an underscore attaced to the number that is redundant).
>        I am not able to use column input because the numbers are not
>        in the same column throughout the data set. So, I am using
>        line input.
 
>        What is the best way to read this as a numeric variable
>        wherein I cannot use a column input format
>        Or is it better to convert into numeric variable reading
>        it as a character variable.
 
Ramu,  I'm not sure that you have any options here.  If you cannot use column
input, then I think you will have to read the variable in as a character
variable, get rid of the non-numeric character(s), and then convert the
'cleaned' character variable to a numeric one.
 
If you know that there is *always* going to be a single non-numeric character
suffix, then the 'cleaning' process might be as simple as:
 
   charvar = substr (charvar, 1, length(charvar)-1) ;
 
Otherwise, you could use the compress() function to extract only the numeric
characters, as recently described on SAS-L by Karsten Self.
 
Either way, the character-to-numeric conversion can then be done using the
input() function.
 
Any help?
 
John
 
-----------------------------------------------------
Dr John Whittington
Mediscience Services Ltd.,
Twyford Manor, Twyford, Buckingham, MK18 4EL  UK
voice:  +44 (0) 1296 730225  fax: +44 (0) 1296 738893
email:  johnw@mag-net.co.uk
        100517.3677@compuserve.com