lp1D: one-dimensional local polynomial smoother

In linulysses/mcfda: Mean and Covariance Estimation for Functional Data Analysis

Description

one-dimensional local polynomial smoother

Usage

lp1D(
  x,
  y,
  h = NULL,
  newx = x,
  degree = 1,
  weight = NULL,
  kernel = "epanechnikov",
  sorted = F
)

Arguments

x	a vector of observed values of the predictor
y	a vector of observed values of the response
h	the bandwidth, either a numeric or NULL; if NULL, the automatically selected by five-fold CV
newx	a vector of new values of the predictor; set to x by default
degree	the degree of local polynomials, nonnegative integer
weight	a vector of real numbers obtaining the weight for each observations; if NULL, then equal weight
kernel	a kernel from the package locpol, supported are 'epanechnikov', "rectangular", "triangular", "quartic", "triweight", "tricube", "cosine", "gauss", "logistic", "sigmoid" and "silverman"
sorted	indicator of whether the array x and newx are sorted in increasing order

Value

a list of the following elements

h	the selected bandwidth or input bandwidth
x	the original input
y	the original input
weight	the original input
kernel	the original input
degree	the original input
yhat	the estimated response at newx

linulysses/mcfda documentation built on Jan. 17, 2021, 8:53 a.m.