em.algo.inner: Inner part of EM-algorithm for QR with measurement error

In bcallaway11/qrme: Quantile Regression with Measurement Error

Description

Does the heavy-lifting of the EM-algorithm for QR with measurment error

Usage

em.algo.inner(
  formla,
  data,
  betmat,
  tau,
  m = 1,
  pi = 1,
  mu = 0,
  sig = 1,
  simstep = "MH",
  iters = 400,
  burnin = 200,
  drawsd = 4,
  cl = 1,
  messages = FALSE
)

Arguments

formla	y ~ x
data	a data.frame that contains y and x
betmat	LxK matrix of parameter values with L the number of quantiles and K the dimension of the covariates
tau	An L-vector indicating which quantiles have been estimated
m	The number of components of the mixture distribution for the measurement error
pi	The probability of each mixture component (should have length equal to m)
mu	The mean of each mixture component (should have length equal to m)
sig	The standard deviation of each mixture component (should have length equal to m)
simstep	Whether to use MH in EM algorithm or importance sampling in EM algorithm. "MH" for MH, and "ImpSamp" for importance sampling. Default is MH.
iters	How many iterations to use in the simulation step (default is 400)
burnin	How many iterations to drop in the simulation step (default is 200)
drawsd	The starting standard deviation for the measurement error term.
cl	The numbe of clusters to use for parallel computation (default is 1 so that computation is not done in parallel)
messages	Whether or not to report details of estimation procedure (default is FALSE)

Value

A list of QR parameters and parameters for mixture of normals for the measurement error term

bcallaway11/qrme documentation built on June 30, 2021, 12:52 p.m.