Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

CHAID w/continuous variables

39 views
Skip to first unread message

bjj...@my-dejanews.com

unread,
Jun 22, 1998, 3:00:00 AM6/22/98
to

My knowledge of SPSS is nearly a decade out of date. Can someone tell me
whether one can use CONTINUOUS (interval or ratio-level scaled) variables as
independent variables in an SPSS-CHAID tree? Specifically, I have a binary
outcome (Dependent variable) I am looking to classify using a bundle of both
nominal and continuous independent variables. I'd like to see the tree
contain some of these continuous variables split at some optimal "cut point"
should they help classify the outcome variable. However, my research vendor
claims this isn't possible with SPSS-CHAID, that SPSS's CHAID product works
only with nominal independent variables.

My experience with CART and prior expectation of SPSS suggests SPSS can indeed
handle both types. Is this not true?

Much, much thanks in advance.

Brent Johnson,Ph.D.
Intel corp market research
brent....@intel.com

-----== Posted via Deja News, The Leader in Internet Discussion ==-----
http://www.dejanews.com/ Now offering spam-free web-based newsreading

Doug Hoy

unread,
Jun 24, 1998, 3:00:00 AM6/24/98
to

On Mon, 22 Jun 1998 13:19:18 GMT, bjj...@my-dejanews.com wrote:


>nominal and continuous independent variables. I'd like to see the tree
>contain some of these continuous variables split at some optimal "cut point"
>should they help classify the outcome variable. However, my research vendor
>claims this isn't possible with SPSS-CHAID, that SPSS's CHAID product works
>only with nominal independent variables.

Don't waste your time with SPSS CHAID. Get KnowledgeSeeker from
Angoss.

*************************************************************
Doug Hoy Evaluation National Museum of
dh...@nmstc.ca & Science & Technology
(613)998-6863v Research P.O.Box 9724, Station T
(613)990-3654f Ottawa K1G 5A3 CANADA
**Opinions expressed are not necessarily shared by the NMSTC**
*************************************************************

0 new messages