Date:         Wed, 29 Sep 2010 10:56:30 -0400
Reply-To:     Arthur Tabachneck <art297@NETSCAPE.NET>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Arthur Tabachneck <art297@NETSCAPE.NET>
Subject:      Re: proc means, class processing vs by processing
Comments: To: Bbser 2009 <bbser2009@GMAIL.COM>

Max,

Since no one has responded, as yet, I'll start the discussion.

First, hopefully someone will tell you what the correct test answer is,
even if it is wrong!

I would think that the correct answer will depend upon which version of
SAS one is running, how many co-processors one has, and how one defines
efficiency.

It's been a long time since I've attempted any time comparisons, as there
are simply too many conflicting factors involved to be able to know
whether one's test is actually valid.

In all of my tests (running 9.1.3 m3 on a 4-processor Windows2003 server),
and looking at real time (rather than cpu time), 'class' actually
outperformed 'by' in all cases EXCEPT when one increased the number of by
variables.

Art
-------
On Wed, 29 Sep 2010 09:46:33 -0400, bbser2009 <bbser2009@GMAIL.COM> wrote:

>I do not quit understand what this passage says about class processing vs
by
>processing in the proc means procedure, especially the last line:
>
>      Because it doesn't require a sorting step, the CLASS statement is
>easier to use than the BY statement.
>      However, BY group processing can be more efficient when you are
>categorizing data that includes many variables.
>
>Sure, I understand that sorted data will be processed more efficiently.
>
>But the last line above seemly conveys a different meanings.
>It reads like the number of the variables in the data set matters about
the
>efficiency.
>
>I am wondering, the variables mentioned here should be understood as
>BY-variables or all the variables in the data set?
>
>Can you say something about this? Thanks a lot in advance.
>
>Regards, Max