|
Margaret,
I'm sure that someone will eventually offer a more elegant solution but,
if you have to get this done before you can celebrate the holidays,
something like the following should at least come close to your desired
solution:
data have;
input A B C D E;
id=_n_;
cards;
25 32 65 32 89
65 32 47 89 96
41 52 63 96 45
12 45 65 98 78
45 56 89 78 63
45 89 56 89 56
56 89 56 45 78
45 56 56 78 89
12 45 98 65 78
12 12 32 35 56
;
run;
proc transpose data=have out=need;
id id;
run;
proc rank data=need out=almost_want;
var _1-_10;
ranks test1-test10;
run;
proc transpose data=almost_want out=almost_want2;
run;
data almost_want3;
set almost_want2;
name=compress(_name_,'_ test');
run;
proc sort data=almost_want3;
by name;
run;
data want;
set almost_want3;
array ranks(5);
array ranks(5);
retain values ranks;
if mod(_n_,2) eq 1 then do;
values(1)=A;
values(2)=B;
values(3)=C;
values(4)=D;
values(5)=E;
end;
else do;
ranks(1)=A;
ranks(2)=B;
ranks(3)=C;
ranks(4)=D;
ranks(5)=E;
output;
end;
run;
%macro doanalysis;
%do i=1 %to 5;
title "Variable" &i.;
proc freq data=want (where=(ranks&i. eq 1));
tables values&i.;
run;
%end;
%mend;
%doanalysis
HTH,
Art
----------
On Sat, 30 Dec 2006 18:26:05 -0500, Margaret Asmani
<margaretasmani@HOTMAIL.CO.UK> wrote:
>Thanks very much everybody for the very useful suggestions.
>
>I use Florio's code as it's convenient not having to list the variables
>again considering that I have a large number. A follow-up question: is it
>possible to compute frequencies only for 1, i.e. the lowest values only?
>
>Happy New Year!!
>
>Margaret
| 
