Hcov {refreg}

Kernel bandwidth selection method based on bivariate density contours coverage

Description

This function implements a method for estimating bivariate kernel bandwidth based on data covarage. The method starts with the plug in estimate (which usually overfits the data), and then increase this bandwidth value until the desired coverage is obtained. Region coverage is evaluated in an out sample design, using a k fold cross validation scheme.

Usage

Hcov(Y, shape = seq(1, 10, 0.5), k = 20, tau = 0.9, display_plot = TRUE)

Arguments

Y

A matrix containing bivariate data values.

shape

A sequence of values which controls plug in estimator increasing.

k

A number indicating k fold cross validations to be performed.

tau

The desired region coverage

display_plot

A logical indicating if a plot must be displaying, during the function estimation process, summarizing the results.

Value

This function return a diagonal kernel bandwidth matrix.