| Library: | gam |
| See also: | intest |
| Macro: | pcad | |
| Description: | pcad estimates the additive components, the significant directions and the regression on principal components |
| Usage: | {jhat,g,mhat} = pcad(x,xg,y,h,bn) | |
| Input: | ||
| x | n x d matrix, the design | |
| xg | ng x d matrix, the points where we want to estimate | |
| y | n x 1 matrix, the response | |
| h | d x 1 matrix or scalar, chosen bandwidth | |
| bn | scalar, threshold for choosing significant directions | |
| Output: | ||
| jhat | q x 1 matrix, the set of significant directions | |
| g | ng x q matrix, additive functions on principal components | |
| mhat | ng x 1 matrix, estimate of regression using the significant functions | |
library("gam")
n = 100
v =uniform(n,4)
x =v[,2:4]
y =x[,1]^2+0.1*x[,2]+normal(n)
h =0.5
bn=0.02
gest=pcad(x,x,y,h,bn)
gest.jhat
gest.g
gest.mhat
The significant directions and the significant function, see Haerdle and Tsybakov "Additive Nonparametric Regression on Principal Components", J. Nonparametric Statist. (1994)157-84.
| Library: | gam |
| See also: | intest |