gom_bayes: Bayesian Grade of Membership Mixture Model

In epopea/gom: Grade of Membership Mixture Models

Bayesian Grade of Membership Mixture Model

Description

This function takes a data object and creates a joint posterior distribution of pure type probabilities, which characterize a small set of extreme profiles, along with the posterior distribution of the grade of membership estimates and the posterior distribution of the Dirichlet distribution hyperparameters.

Usage

gom_bayes(
  data,
  ntypes = 2,
  alpha = "",
  burnin = 250,
  ngibbs = 250,
  omega = 50,
  eta = 10,
  tau = 2,
  beta = 2,
  gomscores = "g"
)

Arguments

data	A data frame with the categorical variables to be used be the model.
ntypes	An integer indicating the number of pure type probabilities to be estimated.
alpha	An array with the pure type probabilities. If specified, the model will not estimate alpha and will only use the array provided by user.
burnin	Number of iterations for the Markov chain achieve a stationary distribution.
ngibbs	Number of iterations after the Markov chain achieved a stationary distribution.
omega	The tuning parameter for the Metropolis–Hastings step.
eta	The tuning parameter of the conditional Dirichlet distribution of Xi.
tau	The shape parameter of the prior Gamma distribution for alpha.
beta	The inverse scale parameter of the prior Gamma distribution for alpha.
gomscores	Prefix for the gamma column names.

Value

An object of class gom_bayes with the posterior distributions of gamma, lambda, Xi, and alpha.

Examples

data <- data.frame(x1 = round(stats::runif(n = 50, 1, 2), 0),
                   x2 = round(stats::runif(n = 50, 1, 3), 0),
                   x3 = round(stats::runif(n = 50, 1, 4), 0))

gom::gom_bayes(data, ntypes = 2, ngibbs = 250, burnin = 250)

epopea/gom documentation built on March 1, 2023, 1:54 a.m.