sampling_Omega_ij: MCMC sampling of parameter "Omega_{i,j}" in the...

In BNPMIXcluster: Bayesian Nonparametric Model for Clustering with Mixed Scale Variables

Description

Generates a sample from the the posterior distribution of the (i,j) element of the Ω matrix in the mixdpcluster model for bayesian clustering. The simulation is done via Metropolis-Hastings method.

Usage

sampling_Omega_ij(
  n = 1,
  Omega.ini,
  i,
  j,
  delta = 4,
  Z,
  mu_Z,
  Lambda,
  sampling_prob,
  n.burn = 0,
  n.thin = 0,
  max.time = Inf,
  verbose = F,
  USING_CPP = USING_CPP
)

Arguments

n	number of simulations to be generated
Omega.ini	matrix Ω with an initialization value for Ω_{i,j}.
i	indicates the row for Ω_{i,j}
j	indicates the column for Ω_{i,j}
delta	defines the maximum jump on each iteration of the MCMC as 1/delta of the feasible interval for Ω_{i,j}
Z	Latent continuous variables
mu_Z	Mean of the continuous latent variables
Lambda	Standard deviations in the decomposition of the covariance matrix
sampling_prob	sampling probabilities for each observation, for complex surveys.
n.burn	number of iterations in the simulation considered in the burn-in period.
n.thin	number of iterations discarded between two simulated values (for thinning of the MCMC chain).
max.time	maximum allowed time for the simulation process. The function returns Error if exceeded.
verbose	if T, the function reports extra information on progress.

Value

A list with two elements:

$omega_ij.chain

A numeric vector with the simulated values from the posterior distribution of Ω_{i,j} .

$accept.indic

A logical vector indicating whether or not omega_ij.prop was accepted.

References

Carmona C., Nieto-Barajas L., Canale A. (2017). Model based approach for household clustering with mixed scale variables.

BNPMIXcluster documentation built on Nov. 30, 2020, 5:07 p.m.