em.algo.inner: Inner part of EM-algorithm for QR with measurement error
In bcallaway11/qrme: Quantile Regression with Measurement Error

em.algo.inner

R Documentation

Inner part of EM-algorithm for QR 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)