Date:         Tue, 9 Feb 1999 17:55:23 +0000
Reply-To:     John Whittington <medisci@POWERNET.COM>
Sender:       "SAS(r) Discussion" <SAS-L@UGA.CC.UGA.EDU>
From:         John Whittington <medisci@POWERNET.COM>
Subject:      Re: Response about Output Window Full
Comments: To: "Berryhill, Timothy" <TWB2@PGE.COM>
Content-Type: text/plain; charset="us-ascii"

At 08:54 09/02/99 -0800, Berryhill, Timothy wrote:
 
>The last time I tried to run a regression with subjects who might be
>distinctly different in their behavior, rather than run many regressions I
>used a BY statement.  This may not be appropriate in this example, but it
>sure sounds like it to me.
 
Tim - that sounds like a close variant of what I posted earlier today - and,
from what he's said, I suspect that probably *is* what Jie Qin is doing.  A
BY statement does, of course, still result in 'many separate regressions
being undertaken' - but the performance advantage is between this (in a
single PROC step) and having thousands of separate PROC runs!
 
Of course, there might actually be an argument for doing it all in one
glorious great analysis!
 
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
----------------------------------------------------------------