latent_hellinger: Hellinger distance for the latent variables of the estimation...

In brsantos/baquantreg: Bayesian quantile regression methods

Hellinger distance for the latent variables of the estimation process.

Description

Returns the Hellinger distance for the latent variables, which are used in estimation process, and it could indicate possible outlying observations.

Usage

latent_hellinger(
  object,
  burnin = 50,
  plot_div = TRUE,
  scales.free = FALSE,
  all.obs = TRUE,
  obs = 1
)

Arguments

object	This is an object of the class "bqr", produced by a call to the bayesQR function.
burnin	Initial part of the chain, which is to be discarded. Default value is 50.
plot_div	If TRUE, the function prints the plot with all probabilities. Default is set to TRUE.
scales.free	If FALSE, the default, then all plots will use the same y scale. If TRUE, for each tau the plot will use the best possible scale in order to visualize the probability information for all observations.
all.obs	if TRUE, calculates KL divergence for all observations
obs	if all.obs is FALSE, specifies the observation to calculate the KL divergence.

Value

Prints a plot of the posterior probability of being an outlier for all observations and returns a data.frame with all values of the probabilities.

brsantos/baquantreg documentation built on July 30, 2024, 1:59 p.m.