Date:         Mon, 20 Sep 1999 14:19:18 +0100
Reply-To:     John Whittington <medisci@POWERNET.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         John Whittington <medisci@POWERNET.COM>
Subject:      Re: Random Selection:  Anything More Elegant
Comments: To: F a b r i z i o <Fabrizio1@USA.NET>
In-Reply-To:  <199909200808.JAA05984@mail-relay.power.net.uk>
Content-Type: text/plain; charset="us-ascii"

At 09:59 20/09/99 +0200, F a b r i z i o wrote:

>Hi John, your algorithm is definitely very nice, I only would
>suggest you replace the references to 'n' by references to
>'totalobs'. At least, I think that is what you need. Am I right?
>('totalobs' is currently uninitialized, and 'n' is not referenced
> at all, hence my conclusion)

Sorry, yes - I converted the 'n' in my usual code to 'totalobs', to be
consistent with the code others have posted - but missed a couple of the
occurrences of 'n.!  The code should have read:

data just20 (drop = needed totalobs);
   retain needed 20 totalobs;
   if _n_=1 then totalobs = total;
   do i = 1 to total ;
      if ranuni(0) le needed/totalobs then do;
         set one nobs = total point = i ;
         needed = needed - 1;
         output ;
      end;
      totalobs = totalobs - 1 ;
      if needed = 0 then stop ;
   end ;
run ;

Sorry for the oversight;  I hope it didn't cause anyone any confusion.

Kind Regards


John

----------------------------------------------------------------
Dr John Whittington,        Voice:    +44 (0) 1296 730225
Mediscience Services        Fax:      +44 (0) 1296 738893
Twyford Manor, Twyford,     E-mail:   medisci@powernet.com
Buckingham  MK18 4EL, UK              mediscience@compuserve.com
----------------------------------------------------------------