zitobitQR: Bayesian quantile regression model with a discrete component...

In brsantos/baquantreg: Bayesian quantile regression methods

Bayesian quantile regression model with a discrete component at zero

Description

This function estimates a bayesian quantile regression model with a discrete component at zero, where all zero observations are assumed to distributed according to a mixed discrete-continuous distribution.

Usage

zitobitQR(
  formula,
  tau = 0.5,
  data,
  itNum,
  thin = 1,
  betaValue = NULL,
  sigmaValue = 1,
  gammaValue = NULL,
  sigmaGamma = 0.5,
  link = 1,
  priorVar = 100,
  refresh = 100,
  quiet = FALSE,
  burnin = 50
)

Arguments

formula	a formula object, with the response on the left of a ~ operator, and the terms, separated by + operators, on the right.
tau	Quantile of interest.
data	a data.frame from which to find the variables defined in the formula
itNum	Number of iterations.
thin	Thinning parameter.
betaValue	Initial values for the parameter beta for the continuous part.
sigmaValue	Initial value for the scale parameter.
gammaValue	Initial value for the parameter gamma of the discrete part.
sigmaGamma	Tuning parameter for the Metropolis-Hastings step.
link	Integer defining the link function used for the probability model. Default is 1. for the logit link function.
priorVar	Value that multiplies a identity matrix in the elicition process of the prior variance of the regression parameters.
refresh	Interval between printing a message during the iteration process. Default is set to 100.
quiet	Logical. If FALSE (default) it will print messages depending on the refresh parameter to show that the Markov chain is updating. If TRUE it will not print messages during the iteration process.
burnin	Size of the burnin only for the indicator variable of the censoring mechanism. For all other chains, this number will not be used. Default value is 50.

Value

A list with the chains of all parameters of interest.

References

Santos and Bolfarine (2015) - Bayesian quantile regression analysis for continuous data with a discrete component at zero. Preprint. http://arxiv.org/abs/1511.05925

Examples

## Not run: 
set.seed(1)
data("BrazilDurableGoods")
# Change the number of iterations for better results.
model <- zitobitQR(expenditure ~ age + education, tau=0.5,
                   data=BrazilDurableGoods, itNum=100,
                   sigmaGamma=0.10, refresh=20)

## End(Not run)

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