Date:         Thu, 16 Sep 2010 13:16:36 -0400
Reply-To:     oloolo <dynamicpanel@YAHOO.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         oloolo <dynamicpanel@YAHOO.COM>
Subject:      Re: Recoding based on frequencies

hash is the way to do:

data one;
  input id (v1-v5) ($);
cards;
1 a a a . a
2 a b b . a
3 a c c . a
4 b c d . a
5 b c e . a
6 c c . . a
;
run;

data two;
     declare hash h();
  h.defineKey('varname', 'varval');
  h.defineData('count');
  h.defineDone();
  length id 3;

  array _v{*} $ v1-v5;
  do until (eof1);
     set one  end=eof1;
  do j=1 to dim(_v);
     varname=vname(_v[j]); varval=vvalue(_v[j]);
           if h.find()=0 then do; count+1; rc=h.replace(); end;
     else do; count=1; rc=h.add(); end;
  end;
  end;

  do until (eof2);
     set one end=eof2;
  do j=1 to dim(_v);
     varname=vname(_v[j]); varval=vvalue(_v[j]);
     if h.find()=0 then _v[j]=count;
     else _v[j]='MISS';
  end;
  output;
  drop rc  varname varval j  count;
  end;
run;



On Thu, 16 Sep 2010 09:55:57 -0400, Chang Chung <chang_y_chung@HOTMAIL.COM>
wrote:

>Hi,
>Saw this interesting question posted somewhere else. I tried, but could not
>come up with a neat solution. Can you? Thanks.
>
>"I have a series of categorical variables that I would like to recode based
>on their frequency/count. [...] So, for example, if I had a series of
>records in the variable being a, a, a, b, b, c, I would like to recode my
>variable so that 'a ' (having the highest count) would be coded as 3 and
>'c' (having the lowest count) would be coded as 1. Since I have a series of
>variables it would be hard to recode them manually so was wondering whether
>there was a command to easily do this."
>
>Below is a test data I made up.
>
>/* test data */
>data one;
>  input id (v1-v5) ($);
>cards;
>1 a a a . a
>2 a b b . a
>3 a c c . a
>4 b c d . a
>5 b c e . a
>6 c c . . a
>;
>run;