Date:         Thu, 28 Nov 2002 19:29:12 -0800
Reply-To:     Bruce Bradbury <BruceBrad@INAME.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Bruce Bradbury <BruceBrad@INAME.COM>
Organization: http://groups.google.com/
Subject:      Most efficient way to subset a large dataset
Content-Type: text/plain; charset=ISO-8859-1

What is the most efficient way to subset a large dataset using records
from another dataset as keys?

Eg dataset SMALL contains a variable ID and is sorted by ID (one
record per ID). Dataset LARGE contains ID plus other variables, and is
also sorted by ID. It is very large: millions of observations and 100
variables. It contains multiple records for each ID. I want to extract
all records from LARGE if they match a record in SMALL. I know I can
do this by using

Data EXTRACT;
merge small (in=insmall) large;
by ID;
if insmall;
run;

Is there are more efficient way of doing this? Eg by using an index
and point operations? Is proc sql likely to be faster?

Any comments welcome.
Bruce Bradbury
[please reply via group]