Hi Joe and ohters, I want to split one huge data set into two parts randomly by a given variable say  by age  in the following data set.

      Id    Age

1          26

2          35

3          26

4          27

5          35

6          24

7          21

8          20

9          23

10       19

The two data sets should have approximately equal number of observations.The output may look like in the following way.

Group A

  Age    id

26       1

26       3

27      4

24      6

19    10

Group B

  Age    id

 35      2

 35      5

21      7

20      8

23      9 As suggested proc surveyselect  may be an appealing approach. But I do not know much about it.  Could do you help here? Thanks in advance  

--- On Tue, 4/28/09, Joe Matise <snoopy369@GMAIL.COM> wrote: From: Joe Matise <snoopy369@GMAIL.COM> Subject: Re: Split data To: SAS-L@LISTSERV.UGA.EDU Date: Tuesday, April 28, 2009, 6:51 AM

You could look at PROC SURVEYSELECT, and select out half of the records... or sort by age and then have an alternating variable (say, dataset=ifn(ranuni(7)<0.5,1,2); ) which will get you approximately even distribution but not exact (perhaps off by 1).

-Joe

On Tue, Apr 28, 2009 at 6:58 AM, Val Krem <valkrem@yahoo.com> wrote:

> Dear SAS users' > > I want to split one huge data set into two parts randomly by > a given variable say by age in the following data set. > > > > Id Age > > 1 > 26 > > 2 > 35 > > 3 > 26 > > 4 > 27 > > 5 > 35 > > 6 > 24 > > 7 > 21 > > 8 > 20 > > 9 > 23 > > 10 > 19 > > > > The two data sets should have approximately > equal number of observations. Remember there are eight unique age groups. > > > > How do I do that? > > > > > > > > > >