Library: | gam |
See also: | intest pcad |
Quantlet: | intest1 | |
Description: | estimation of the univariate additive functions in a separable additive model using Nad.Wat. |
Usage: | gest = intest1(x,y,xg,h,g) | |
Input: | ||
x | n x d matrix , the observed explanatory variable where the directions of interest have to be the first q columns | |
y | n x p matrix , the observed response variables | |
xg | m x q matrix , the grid with m points in each of the q directions of interest | |
h | q x 1 or 1 x 1 matrix , chosen bandwidth for the directions of interest | |
g | d x 1 or 1 x 1 matrix , chosen bandwidth for the directions not of interest | |
Output: | ||
gest | m x q x p matrix, containing the marginal integration estimators |
library("gam") n = 150 x = uniform(n,4)*4-2 g1 = 2*x[,1] g2 = x[,2]^2 - 4/3 g3 = exp(x[,3]) g4 = sin(1.5*x[,4]) eps = normal(n,1) * sqrt(0.5) y = g1 + g2 + g3 + g4 + eps xg = grid(-1.8,0.2,19) xg = xg~xg h = #(1.0, 0.75) ; we are interested in g = #(1.3, 1.0, 1.5, 1.5) ; the shape of g1, g2 gest = intest1(x,y,xg,h,g) bild = createdisplay(1,2) dat11 = x[,1]~g1 dat12 = xg[,1]~gest[,1] dat21 = x[,2]~g2 dat22 = xg[,2]~gest[,2] setmaskp(dat12,4,4,8) setmaskp(dat22,4,4,8) setmaskl(dat12,(1:rows(dat12))',4,1,1) setmaskl(dat22,(1:rows(dat22))',4,1,1) show(bild,1,1,dat11,dat12) show(bild,1,2,dat21,dat22)
the marginal integration estimator of the additive functions, using a multidimensional Nadaraya Watson see Tjostheim and Auestad, "Nonparametric Identifi- cation of Nonlinear Time Series: Projections", JASA, (1994)