Date:         Fri, 20 Aug 2004 14:09:51 GMT
Reply-To:     Rusty Shackleford <rs@OVERLOOK.HOMELINUX.NET>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Rusty Shackleford <rs@OVERLOOK.HOMELINUX.NET>
Organization: SBC http://yahoo.sbc.com
Subject:      How to get unique values faster than sorting?

I use proc format to set up mappings from one var, like ssn, to another
var, like age all the time.  Before I can use proc format though, I have
to run some enormous datasets through proc sort nodupkey first and this
takes a while.

Since I'm using formats, I don't need my data sorted, just reduced to
one obs per key.  Is there a faster way computationally to do this
than:

    data dd;
        set bigdata (keep=ssn age);
    run;

    proc sort nodupkey data=dd;
        by ssn;

    /* proc format ssn -> age here. */
    proc format...

Thanks!

--
Give and take free stuff: http://freecycle.org