Date:         Sun, 29 Apr 2001 17:33:07 -0500
Reply-To:     seymour.d.douglas@ACCENTURE.COM
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Seymour Douglas <seymour.d.douglas@ACCENTURE.COM>
Subject:      Re: off topic- prediction model for income estimation
Comments: To: isaac wong <duckchai@NETVIGATOR.COM>
Content-type: text/plain; charset=us-ascii

Isaac,
           I think treating income as a categorical variable is a so-so
idea. The only model that would satisfy the ordinal nature of the
categoreis is an ordered probit.

Accenture is the new name for Andersen Consulting as of January 1, 2001.
Our web address is http://www.accenture.com


isaac wong <duckchai@NETVIGATOR.COM> (Mailed by: "SAS(r) Discussion"
<SAS-L@LISTSERV.UGA.EDU>)
04/30/2001 12:49 AM MST
Please respond to isaac wong <duckchai@NETVIGATOR.COM>

To:   SAS-L@LISTSERV.UGA.EDU
cc:
Subject:  off topic- prediction model for income estimation


Hi all,
    recently i was given a task in a bank to build a predictive model for
income estimation. I  wonder if  treating the income (dependent variable)
as
a catagorical variable (grouping income as predefined level) will be
superior than taking income as a continuious interval variable. I m not
sure
which way will be more accurate for prediction.

    Would u have any suggestion for me. thx

isaac