Generalized Additive Models
Let's assume that we have independent random variables
and
and the
response variable
having the form
The functions
,
are unknown and the random error
is independent with
and
. This situation can be simulated using
XploRe
very well:
n = 100
t = normal(n,2) ; explanatory variable
f1 = - sin(2*t[,1]) ; estimated functions
f2 = t[,2]^2
eps = normal(n,1) * sqrt(0.75) ; error
y = f1 + f2 +eps ; response variable
The data can come from praxis, too.
Our task is to estimate the unknown functions
and
.
This chapter deals with such problems and their solutions. It ought
to demonstrate and to explain how to use
XploRe
for nonparametric regression and data analysis in
generalized additive models (GAM).
It describes all quantlets which belong to the gam
quantlib which
contains all
routines of
XploRe
provided for estimation and testing in
generalized additive models. It also has several links to the
gplm
quantlib for generalized partial linear models
(GPLM) in
XploRe
thus many quantlets which are used in
gam
are fully described in Chapter
but not mentioned here.