Local Linear Regression Smoother Matrix

Compute the Local Linear Regression (LLR) smoother matrix. LLR has better boundary bias properties than Nadaraya-Watson.

Usage

S.LLR(tt, h, Ker = "norm", w = NULL, cv = FALSE)

Arguments

tt

Evaluation points (numeric vector).

h

Bandwidth parameter.

Ker

Kernel function or name. One of "norm", "epa", "tri", "quar", "cos", "unif".

w

Optional weights vector of length n.

cv

Logical. If TRUE, compute leave-one-out cross-validation matrix.

Value

An n x n smoother matrix S.

Examples

tt <- seq(0, 1, length.out = 50)
S <- S.LLR(tt, h = 0.1)