mcmc_bin_metropolis: MCMC for binary regression - Metropolis version
In DouglasMesquita/robit: FLAMES: Flexible Link functions with AsyMptotES

mcmc_bin_metropolis

R Documentation

MCMC for binary regression - Metropolis version

Description

Perform MCMC with Metropolis Hastings algorithm

Usage

mcmc_bin_metropolis(y, X, nsim, burnin, lag, inv_link_f, type, sample_c,
  sample_d, sigma_beta, a_c, b_c, a_d, b_d, a_lambda, b_lambda, var_df,
  var_c, var_d, var_lambda, p_c, p_d, p_prop, p_beta, p_df, p_lambda,
  const, const_beta, const_c, const_d, const_df, const_lambda)

Arguments

`y`	Bernoulli observed values
`X`	Covariate matrix
`nsim`	Sample size required for MCMC
`burnin`	Burn in for MCMC
`lag`	Lag for MCMC
`inv_link_f`	Inverse link function
`type`	"logit", "probit", "cauchit", "robit", "cloglog" or "loglog"
`sample_c`	Should c be sampled?
`sample_d`	Should c be sampled?
`sigma_beta`	Variance of beta prior
`a_c`	Shape1 for c prior
`b_c`	Shape2 for c prior
`a_d`	Shape1 for d prior
`b_d`	Shape2 for d prior
`a_lambda`	Inferior limit for lambda
`b_lambda`	Superior limit for lambda
`var_df`	Variance to sample 1-exp(-df/const)
`var_c`	Variance to sample from c (if sample_d = TRUE) otherwise log(c/(1-c))
`var_d`	Variance to sample d (if sample_c = TRUE) otherwise log(c/(1-c))
`var_lambda`	Variance to sample lambda
`p_c`	To restore c
`p_d`	To restore d
`p_prop`	To restore p
`p_beta`	To restore beta
`p_df`	To restore df
`p_lambda`	To restore lambda
`const`	A constant to help on sampling degrees of freedom \tilde{df} = df/c
`const_beta`	A constant to tunning the acceptance rate (default = 2.38^2)
`const_c`	A constant to tunning the acceptance rate (default = 2.38^2)
`const_d`	A constant to tunning the acceptance rate (default = 2.38^2)
`const_df`	A constant to tunning the acceptance rate (default = 2.38^2)
`const_lambda`	A constant to tunning the acceptance rate (default = 2.38^2)