winsorize data

[danielhu]

Hi, can I get some help on how to winsorize my data. I have total of 8 variables (for example, A-H) with 8000 observations that I have to winsorize. Thanks in advance!

[andreas_lds]

Maybe that old thread has the answer for you: or this one

[JonasE]

Maybe this link will help:

Otherwise look at this solution and you might be able to work from there. It is not the neatest but your question is not so specific. I have used fifth and ninetyfifth percentiles and set those values to missing.

data test;

set sashelp.cars;

keep Horsepower ;

run;

proc univariate data=test noprint;

var Horsepower;

output out=percentiles p5=Horsepower_p5 p95=Horsepower_p95 ;

run;

data want;

if _n_=1 then set percentiles;

set test;

if not (Horsepower_p5 < horsepower < Horsepower_p95) then horsepower=. ;

run;

[Ksharp]

IML code:

Code: Program

data have;
 call streaminit(1234);
 do i=1 to 100;
  a=ceil(rand('uniform')*100);
  b=ceil(rand('uniform')*100);
  output;
 end;
 drop i;
run;

%let low=0.05;
%let high=0.95;

proc iml;
use have;
read all var _num_ into x[c=vname];
close have;

call qntl(q,x,{&low , &high});
do i=1 to ncol(x);
 x[loc(x[,i]<q[1,i]),i]=.;
 x[loc(x[,i]>q[2,i]),i]=.;
end;

create want from x[c=vname];
append from x;
close want;
quit;