Date:         Fri, 12 Oct 2001 13:50:50 -0700
Reply-To:     Cassell.David@EPAMAIL.EPA.GOV
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         "David L. Cassell" <Cassell.David@EPAMAIL.EPA.GOV>
Subject:      Re: PROC UNIVARIATE
Content-type: text/plain; charset=us-ascii

Rich wrote [in part]:
> I need to find the maximum value of a variable within a certain class
> of another variable. . . . . .
>
> The code works but I keep getting the following WARNING message in the
> LOG file. As a result I need to clean the LOG window from time to time
> in order to keep the program running.
>
>   WARNING: The range of variable TERMTIME is too small relative to its
>   mean, numerical inaccuracies may occur. Recode if possible.

Well, it's hard to give you an optimal plan when you haven't said why
you're
doing this.  But if you are doing it so often that you are filling up
your log
window, then you're probably not doing the best approach.

If you need multiple maxima of TERMTIME by PHASE, then perhaps you
should:
[1] not use PROC UNIVARIATE at all, but something a little faster like
PROC SUMMARY;
[2] do all the maxima at once, using a BY PHASE; statement in the PROC
and save all
    the results in a new dataset;
[3] then you can access the outputted dataset as needed, perhaps merging
it with
    your original dataset to get all maxima at once, or using the
dataset of maxima
    as a table lookup.

And finally, you really ought to think about why your values are so
close to
identical within each value of PHASE.  Are they supposed to be that way?
If so,
then is there any point in pulling out the max within a value of PHASE
if they all
look to be about the same value?

HTH,
David
--
David Cassell, CSC
Cassell.David@epa.gov
Senior computing specialist
mathematical statistician