Keywords - Function groups - @ A B C D E F G H I J K L M N O P Q R S T U V W X Y Z

Library: gam
See also: intest intestpl gintest gintestpl backfit

Quantlet: sptest
Description: Additive component analysis in additive separable models using wavelet estimation. An additive component can be tested against a given polynomial form with degree p, e.g. when p is set to zero we test for significant influence of that component. The procedure is presented in Haerdle, Sperlich, Spokoiny (2000) but implemented without the "t_{j,alpha}" correction.

Usage: pv = sptest(x,y,p{,levels})
Input:
x n x p matrix , the observed explanatory variable where the discrete variables have to be placed into the last (dis) columns
y n x 1 matrix , the observed response variable
p scalar , the polynomial degree of the Hypothesis
levels dis x 1 vector of integrers < 4, wavelet levels for the components of the discrete variables; max = 3
Output:
pv p-value of the hypothesis

Example:
library("gam")
n   = 100
randomize(1234)
x   = normal(n,3)
eps = normal(n,1) * sqrt(0.8)
y   = sin(2*x[,1]) + x[,2]^2 + 2*x[,3] +eps
p   = 1
pv = sptest(x,y,p)

Result:
table of results



Author: S. Sperlich, 20000622
(C) MD*TECH Method and Data Technologies, 05.02.2006