Help using Base SAS procedures

winsorization

Accepted Solution Solved
Reply
Contributor
Posts: 72
Accepted Solution

winsorization

I want to winsorize the attached data at 1% and 99% winsorization on the basis of country and family and non family firms. like in this format. can anyone help me?

Country Kor

                                         Book deb ratio
                               N    mean   median  max  min
 Family firms
Non-family firms

Accepted Solutions
Solution
‎06-17-2016 04:07 AM
Super User
Posts: 9,656

Re: winsorization

I used your original 22.xls file .

 

 

proc import datafile='/folders/myfolders/22.xls' out=have replace dbms=xls;mixed=yes;run;
proc format;
value $ fmt
 '0'='Non-family firms'
 '1'='Family firms';
run;
proc sort data=have;by count fam;run;
proc summary data=have ;
by count fam;
var book_debt_ratio;
output out=temp(drop=_freq_ _type_) p1= p99= /autoname;
run;
data want;
 merge have temp;
 by count fam;
 if book_debt_ratio lt book_debt_ratio_P1 then book_debt_ratio=book_debt_ratio_P1;
 if book_debt_ratio gt book_debt_ratio_P99 then book_debt_ratio=book_debt_ratio_P99;
run;

options nobyline;
title  "Country #byval1";
proc tabulate data=want;
by count;
var book_debt_ratio;
class Fam;
format Fam $fmt.;
table Fam=' ',book_debt_ratio*(N    mean   median  max  min);
run;

View solution in original post


All Replies
Contributor
Posts: 62

Re: winsorization

Have you had a look at the discussion here?

 

https://communities.sas.com/t5/SAS-Procedures/How-to-winsorize/td-p/214976

 

The discussion includes the difference between winsorising and trimming.

Norman.
SAS 9.4 (TS1M0) X64_7PRO WIN 6.1.7601 Service Pack 1 Workstation

Solution
‎06-17-2016 04:07 AM
Super User
Posts: 9,656

Re: winsorization

I used your original 22.xls file .

 

 

proc import datafile='/folders/myfolders/22.xls' out=have replace dbms=xls;mixed=yes;run;
proc format;
value $ fmt
 '0'='Non-family firms'
 '1'='Family firms';
run;
proc sort data=have;by count fam;run;
proc summary data=have ;
by count fam;
var book_debt_ratio;
output out=temp(drop=_freq_ _type_) p1= p99= /autoname;
run;
data want;
 merge have temp;
 by count fam;
 if book_debt_ratio lt book_debt_ratio_P1 then book_debt_ratio=book_debt_ratio_P1;
 if book_debt_ratio gt book_debt_ratio_P99 then book_debt_ratio=book_debt_ratio_P99;
run;

options nobyline;
title  "Country #byval1";
proc tabulate data=want;
by count;
var book_debt_ratio;
class Fam;
format Fam $fmt.;
table Fam=' ',book_debt_ratio*(N    mean   median  max  min);
run;
SAS Super FREQ
Posts: 3,473

Re: winsorization

See the discussion and links in the article "How to Winsorize data in SAS."

☑ This topic is SOLVED.

Need further help from the community? Please ask a new question.

Discussion stats
  • 3 replies
  • 322 views
  • 3 likes
  • 4 in conversation