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 |
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)
table of results