Date: Sun, 21 Sep 1997 21:27:21 -0400
Reply-To: "Muhlbaier, Lawrence" <muhlb001@ONYX.MC.DUKE.EDU>
Sender: "SAS(r) Discussion" <SAS-L@UGA.CC.UGA.EDU>
From: "Muhlbaier, Lawrence" <muhlb001@ONYX.MC.DUKE.EDU>
Subject: Re: spline function
Content-Type: text/plain
The short answer is yes.
Spline functions are just specially formatted linear combinations and
can be used as another piece of a linear combination as you have
described. Note also that spline functions are applied to
continuous-valued X's and really don't make any sense for binary
variables like marital status (technically you could construct a spline
function on a binary variable, but it would degenerate back into a
binary variable).
For another time, there is a very readable article about splines by
Stone and Koo in, I believe, the American Statistician (I don't have the
exact reference handy).
Doc Muhlbaier
> ----------
> From: Jie Qin[SMTP:c674010@SHOWME.MISSOURI.EDU]
> Sent: Sunday, September 21, 1997 9:00 PM
> To: SAS-L@UGA.CC.UGA.EDU
> Subject: spline function
>
> Hi! everybody. My background is not statistics, but I need to use
> spline
> function to finish a term paper. I have a basic question for spline
> function. If I represent the spline function by Y=S(X), then can I
> incorporate a dummy variable Z additively in the form Y=S(X)+kZ?
> Should
> this variable Z be designed to be another form of spline function?
> Supposed Z is represented marriage status (single or married).
>
> Your assistance will be greatly appreciated.
>
> Jie Qin
> 1720 East Broadway Rd.
> Apt. D
> Columbia, MO 65201
> Tel: (573) 817-3170
> http://www.missouri.edu/~c674010
>
| 
