Date:         Fri, 19 Dec 2003 06:40:17 GMT
Reply-To:     "Andrew H. Karp" <sfbay0001@AOL.COMNOPE>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         "Andrew H. Karp" <sfbay0001@AOL.COMNOPE>
Organization: AOL http://www.aol.com
Subject:      Re: function for quartile flags (cut points)

>You should look at PROC RANK.  You can split your data into 10 groups
>for deciles of variable INCOME like this:
>
>proc rank data=yourdata out=rankdata groups=10;
>  var income;
>  run;
>
>This will replace INCOME in the output data set with the ranking values,
>1 to 10.  To do quartiles, just change the first line to GROUPS=4
>instead.
>
>To keep INCOME and build an INCOMERANK variable with the rank-group
>values, create a duplicate variable first, like this:
>
>data yourview/view=yourview;
>  set yourdata;
>  incomerank=income;
>  run;
>
>proc rank data=yourdata out=rankdata groups=10;
>  var incomerank;
>  run;

Let me please add to David's post on this topic.  I prefer to use the RANKS
statement in PROC MEANS, in which you give the names of the variable(s) in the
output data set that hold/store the rankings generated by the PROC.  This
avoids either over-writing the values of the analysis variable with their
ranks, or having to create a separate variable in a data step before applying
PROC RANK.

Here is what I would do:

PROC RANK DATA=yourdata groups=10 our=ranked_data;
var income;
ranks incomerank;
run;

ALL of the variables in "yourdata" will be output to "ranked_data" along with
the new variable INCOMERANK, which holds the deciles.

Hope this helps...

Andrew




Andrew H. Karp
SAS Certified Professional, Version 8
Sierra Information Services, Inc.
19229 Sonoma Hwy.
Sonoma, CA 95476 USA
707/996-7380
SierraInfo@AOL.COM
http://www.SierraInformation.com