Date:         Tue, 27 Aug 2002 09:02:48 -0400
Reply-To:     "Goldman, Brad (AT-Atlanta)" <Brad.Goldman@AUTOTRADER.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         "Goldman, Brad (AT-Atlanta)" <Brad.Goldman@AUTOTRADER.COM>
Subject:      Re: Efficiency advice for nodups
Comments: To: "Dorfman, Paul" <Paul.Dorfman@bcbsfl.com>
Content-Type: text/plain; charset="iso-8859-1"

Thanks to all for the advice.  Lots of good ideas in there (so far I favor
Puddin' Man's Update and Paul's Proc Means approach.)

One part of the specs I wasn't clear on is the structure and sorting of the
datasets.  It would like this:
day 11:
car     date
1       11
2       11
3       11
day 12:
car     date
1       12
3       12
day 13:
car     date
1       13
4       13
Final data should be:
car     date
1       13
2       11
3       12
4       13
This affects some of the solutions, since some of them assumed the combined
dataset would already be sorted by car/date -- that's not neccessarily true
depending on the approach used.

Thanks again to all,
Brad Goldman