| Library: | gam |
| See also: | intest intestpl gintest gintestpl backfit |
| Macro: | 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
| Library: | gam |
| See also: | intest intestpl gintest gintestpl backfit |