nadwat: The Classical Nadaraya-Watson Regression Estimator

In kader: Kernel Adaptive Density Estimation and Regression

Description

In its arguments x and dataX vectorized function to compute the classical Nadaraya-Watson estimator (as it is m_n in eq. (1.1) in Eichner & Stute (2012)).

Usage

nadwat(x, dataX, dataY, K, h)

Arguments

x	Numeric vector with the location(s) at which the Nadaraya-Watson regression estimator is to be computed.
dataX	Numeric vector (X_1, …, X_n) of the x-values from which (together with the pertaining y-values) the estimate is to be computed.
dataY	Numeric vector (Y_1, …, Y_n) of the y-values from which (together with the pertaining x-values) the estimate is to be computed.
K	A kernel function (with vectorized in- & output) to be used for the estimator.
h	Numeric scalar for bandwidth h.

Details

Implementation of the classical Nadaraya-Watson estimator as in eq. (1.1) in Eichner & Stute (2012) at given location(s) in x for data (X_1, Y_1), …, (X_n, Y_n), a kernel function K and a bandwidth h.

Value

A numeric vector of the same length as x.

Examples

require(stats)

 # Regression function: a polynomial of degree 4 with one maximum (or
 # minimum), one point of inflection, and one saddle point.
 # Memo: for p(x) = a * (x - x1) * (x - x2)^3 + b the max. (or min.)
 # is at x = (3*x1 + x2)/4, the point of inflection is at x =
 # (x1 + x2)/2, and the saddle point at x = x2.
m <- function(x, x1 = 0, x2 = 8, a = 0.01, b = 0) {
 a * (x - x1) * (x - x2)^3 + b
 }
 # Note: for m()'s default values a minimum is at x = 2, a point
 # of inflection at x = 4, and a saddle point at x = 8.

n <- 100       # Sample size.
set.seed(42)   # To guarantee reproducibility.
X <- runif(n, min = -3, max = 15)      # X_1, ..., X_n
Y <- m(X) + rnorm(length(X), sd = 5)   # Y_1, ..., Y_n

x <- seq(-3, 15, length = 51)   # Where the Nadaraya-Watson estimator
                                # mn of m shall be computed.
mn <- nadwat(x = x, dataX = X, dataY = Y, K = dnorm, h = n^(-1/5))

plot(x = X, y = Y);   rug(X)
lines(x = x, y = mn, col = "blue")  # The estimator.
curve(m, add = TRUE, col = "red")   # The "truth".

kader documentation built on May 1, 2019, 10:13 p.m.