compKernVals: Compute kernel values.

In locpol: Kernel Local Polynomial Regression

Compute kernel values.

Description

Some R code provided to compute kernel related values.

Usage

	computeRK(kernel, lower=dom(kernel)[[1]], upper=dom(kernel)[[2]],
			subdivisions = 25)
	computeK4(kernel, lower=dom(kernel)[[1]], upper=dom(kernel)[[2]],
			subdivisions = 25)
	computeMu(i, kernel, lower=dom(kernel)[[1]], upper=dom(kernel)[[2]],
			subdivisions = 25)
	computeMu0(kernel, lower=dom(kernel)[[1]], upper=dom(kernel)[[2]],
			subdivisions = 25)
	Kconvol(kernel,lower=dom(kernel)[[1]],upper=dom(kernel)[[2]],
			subdivisions = 25)

Arguments

kernel	Kernel used to perform the estimation, see Kernels
i	Order of kernel moment to compute
lower, upper	Integration limits.
subdivisions	the maximum number of subintervals.

Details

These functions uses function integrate.

Value

A numeric value returning:

computeK4	The fourth order autoconvolution of K.
computeRK	The second order autoconvolution of K.
computeMu0	The integral of K.
computeMu2	The second order moment of K.
computeMu	The i-th order moment of K.
Kconvol	The autoconvolution of K.

These functions are implemented by means of integrate.

Author(s)

Jorge Luis Ojeda Cabrera.

References

Fan, J. and Gijbels, I. Local polynomial modelling and its applications\/. Chapman & Hall, London (1996).

Wand, M.~P. and Jones, M.~C. Kernel smoothing\/. Chapman and Hall Ltd., London (1995).

See Also

RK, Kernel characteristics, integrate.

Examples

	##	Note that lower and upper params are set in the definition to
	##	use 'dom()' function.
	g <- function(kernels)
	{
		mu0 <- sapply(kernels,function(x) computeMu0(x,))
		mu0.ok <- sapply(kernels,mu0K)
		mu2 <- sapply(kernels,function(x) computeMu(2,x))
		mu2.ok <- sapply(kernels,mu2K)
		Rk.ok <- sapply(kernels,RK)
		RK <- sapply(kernels,function(x) computeRK(x))
		K4 <- sapply(kernels,function(x) computeK4(x))
		res <- data.frame(mu0,mu0.ok,mu2,mu2.ok,RK,Rk.ok,K4)
		res
	}
	g(kernels=c(EpaK,gaussK,TriweigK,TrianK))

locpol documentation built on Nov. 29, 2022, 9:05 a.m.