NW.kernel(haerdle)R Documentation

Nadaraya-Watson non-parametric regression

Description

Nadaraya-Watson non-parametric regression

Usage

NW.kernel(x, y, h, kernel=4, points=100, na.handling=0)

Arguments

Required:
x data vector
y data vector
h bandwidth
kernel code for kernel. 1 = uniform, 2 = triangle (ASH), 3 = Epanchenikov, 4 = quartic, 5 = triweight, 6 = Gaussian, 7 = cosinus
points number of points to evaluate curve over the data range
na.handling How to handle 0/0. If 1 forces 0/0 = 0.

Value

list with components grid, m (fitted curve), x and y

References

`Smoothing Techniques with Implementation in S', Wolfgang Haerdle, Springer, 1991

Examples

data(faithful)
# Figure 5.3
plot(faithful$eruptions,faithful$waiting)
nw<-NW.kernel(faithful$eruptions, faithful$waiting, 0.1)
lines(nw$grid, nw$m)
nw<-NW.kernel(faithful$eruptions, faithful$waiting, 0.4)
lines(nw$grid, nw$m, lty=2)
nw<-NW.kernel(faithful$eruptions, faithful$waiting, 0.8)
lines(nw$grid, nw$m, lty=3)

data(dat.reg)
# Figure 5.4
plot(dat.reg)
lines(dat.reg$x, dat.reg$m)
nw<-NW.kernel(dat.reg$x, dat.reg$y, 0.05)
lines(nw$grid, nw$m, lty=2)
nw<-NW.kernel(dat.reg$x, dat.reg$y, 0.10)
lines(nw$grid, nw$m, lty=3)


[Package Contents]