Date:         Fri, 12 Feb 2010 01:12:51 -0800
Reply-To:     xlr82sas <xlr82sas@AOL.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         xlr82sas <xlr82sas@AOL.COM>
Organization: http://groups.google.com
Subject:      Re: Character to Numeric variable conversion
Comments: To: sas-l@uga.edu
Content-Type: text/plain; charset=ISO-8859-1

On Feb 11, 12:22 pm, jim.1s...@YAHOO.COM (Jim Groeneveld) wrote:
> Hi Yaw,
>
> I wrote a macro that 'converts' character variables into numeric ones by
> firstly copying the other type result in another variable that is renamed to
> the original name later. So from a user point of view, calling the macro it
> really looks like type conversion.
>
> Additionally the original variable label, format and informat are retained
> (copied), the format names are just changed ($ removed); I once have to
> interpret and adapt the actual formats themselves, this is still left to the
> user.
>
> Furthermore there may be non-numerics in the character data, like 'NA', 'M',
> or whatever. These character values can be assigned any desired numeric
> values (missings as well) during the macro call.
>
> The macro is available as:http://jim.groeneveld.eu.tf/software/SASmacro/aRecodeN.zip
>
> See alsohttp://listserv.uga.edu/cgi-bin/wa?S2=sas-l&q=aRecodeN&s=&f=&a=2002&b=
>
> Regards - Jim.
> --
> Jim Groeneveld, Netherlands
> Statistician, SAS consultanthttp://jim.groeneveld.eu.tf
>
>
>
> On Tue, 9 Feb 2010 13:12:32 -0800, Yaw <link...@GMAIL.COM> wrote:
> >Dear Community:
>
> >I have this dataset with character variables I want to convert to
> >numeric. The data points are all numeric but for some reason they are
> >being read as character.
>
> >So I tried converting them by using  charvar*1  or  Charvar +0. Proc
> >contents told me they have been converted(into numeric) but when I run
> >a proc means on the variables I see the vars as missing  mean sd  etc
> >and N  of 0.
>
> >Can somebody tell me what's amiss here? Any ideas?  Thanks in advance.
> >Yaw
>
> >Data:
>
> >id     mx
> > 1       .
> > 2       0.53
> > 5         .
> >12    15 .5
> >13       .
> > etc.- Hide quoted text -
>
> - Show quoted text -

You can alwasy use SQL. There also may be a way using aliases in proc
report.


data chr;
   one='1';
   two='2';
   tre='3';
   for='4';
run;

proc sql;
  create
     table num as
  select
      input(one,1.) as one
     ,input(two,1.) as two
     ,input(tre,1.) as tre
     ,input(for,1.) as for
  from
     chr
;quit;