Numerical-MLE: Numerical MLE

In mixlink: Mixture Link Regression

Description

Numerical maximum likelihood estimation for Mixture Link Binomial and Poisson models.

Usage

mle.mixlink.binom(y, m, J, extra.tx = null.tx, var.names = NULL,
  phi.init = NULL)

mle.mixlink.binom.x(y, m, X, J, extra.tx = null.tx, var.names = NULL,
  phi.init = NULL, invlink.mean = plogis)

mle.mixlink.pois(y, J, extra.tx = null.tx, var.names = NULL,
  phi.init = NULL)

mle.mixlink.pois.x(y, X, J, extra.tx = null.tx, var.names = NULL,
  phi.init = NULL, invlink.mean = exp)

Arguments

y	Argument of pdf or cdf.
m	Number of success/failure trials.
J	Number of mixture components to use.
extra.tx	If additional functions of \bm{θ} are to be estimated, they can be specified here. The default null.tx indicates that no extra functions are desired.
var.names	A vector of strings to use for parameter names. The default (NULL) indicates to leave the names at some defaults.
phi.init	Intial value of the unconstrained \bm{φ} parameters. Internally, a transformation \bm{θ} is applied to \bm{φ} to obtain parameters in the correct space. The default (NULL) selects a default initial value.
X	Design matrix for regression case.
invlink.mean	The inverse link function for the mean. Default is plogis for Binomial and exp for Poisson.

Value

A list with MLE results. Can be accessed through the functions confint, print, summary, coef, logLik AIC, BIC, and vcov.

References

Andrew M. Raim, Nagaraj K. Neerchal, and Jorge G. Morel. An Extension of Generalized Linear Models to Finite Mixture Outcomes. arXiv preprint: 1612.03302

Examples

## Not run: 
  n <- 400
  mean.true <- rep(20, n)
  Pi.true <- c(1/5, 4/5)
  kappa.true <- 2
  y <- r.mixlink.pois(n, mean.true, Pi.true, kappa.true)

  mle.out <- mle.mixlink.pois(y, J = 2)
  coef(mle.out)
  print(mle.out)
  confint(mle.out)

## End(Not run)

mixlink documentation built on May 2, 2019, 5:11 a.m.