tvReg-internals: tvReg internal and secondary functions

.tvLM.ciR Documentation

tvReg internal and secondary functions

Description

Listed below are supporting functions for the major methods in tvReg.

Create a list of control pararameters for the tvSURE and tvPLM methods. All control parameters that are not passed to this function are set to default values.

Usage

.tvLM.ci(x, yboot)

.tvSURE.ci(x, yboot)

.kernel(x, bw, tkernel = "Triweight")

.univariatePlot(x, vars = NULL, ylim = NULL, ...)

.tvCov.cv(
  bw,
  x,
  z = NULL,
  cv.block = 0,
  est = c("lc", "ll"),
  tkernel = c("Triweight", "Epa", "Gaussian")
)

.tvFE.cv(
  bw,
  x,
  y,
  z = NULL,
  neq,
  obs,
  cv.block = 0,
  est = c("lc", "ll"),
  tkernel = c("Triweight", "Epa", "Gaussian")
)

.tvIRF(
  x,
  impulse,
  response,
  y.names,
  n.ahead,
  ortho,
  ortho.cov,
  bw.cov,
  cumulative
)

.tvOLS.cv(
  bw,
  x,
  y,
  z = NULL,
  cv.block = 0,
  est = c("lc", "ll"),
  tkernel = c("Triweight", "Epa", "Gaussian"),
  singular.ok = TRUE
)

.tvRE.cv(
  bw,
  x,
  y,
  z = NULL,
  neq,
  obs,
  cv.block = 0,
  est = c("lc", "ll"),
  tkernel = c("Triweight", "Epa", "Gaussian")
)

tvreg.control(maxiter = 100, tol = 1e-05)

Arguments

x

A matrix.

bw

A scalar.

tkernel

A character, either "Triweight" (default), "Epa" or "Gaussian" kernel function.

z

A vector.

est

A character, either "lc" or "ll" for local constant or local linear.

maxiter

maximum number of iterations for the iterative FGLS estimations.

tol

tolerance level indicating when to stop the iteration for the iterative FGLS estimations

Details

.kernel calculates the kernel values of a vector and a given bandwidth

If the estimation is iterative FGLS with maxiter>1, the convergence criterion is

\sqrt{ \frac{ \sum_{i, j} (B_{i,j,g} - B{i, j, g-1})^2 }{ \sum_{i, j} B_{i, j, g-1}^2 }} < \code{tol}

(B_{i, j,g} is the ith, jth coefficient of the gth iteration step).

Value

A numeric vector.

A scalar with the mean squared error.

A list of the above components.


tvReg documentation built on Sept. 1, 2023, 5:07 p.m.