Skip to contents

Plot Gelman-Rubin-Brooks

Source: R/mod_plot_gelman.R
plot_gelman.Rd

The PlotGelman_*() functions generate plots visualising the evolution of the Gelman-Rubin-Brooks shrink factor for different model parameters as the number of iterations increases. These plots help assess whether MCMC chains have converged to a common distribution. Each plot includes: median (solid line) and 97.5th percentile (dashed line) of the shrink factor and a dashed horizontal line at 1.1, representing the common convergence threshold. The primary function for users is plot_gelman(), which internally calls:

  • plot_gelman_alpha(): Plots shrink factor for the Alpha parameter

  • plot_gelman_beta(): Plots shrink factor for the Beta parameters

  • plot_gelman_omega(): Plots shrink factor for the Omega parameter

  • plot_gelman_rho(): Plots shrink factor for the Rho parameter

Usage

plot_gelman(
  path_coda = NULL,
  alpha = TRUE,
  beta = TRUE,
  omega = TRUE,
  rho = TRUE,
  n_omega = 1000L,
  plotting_alpha = 0.25,
  env_file = ".env",
  return_plots = FALSE
)

plot_gelman_alpha(coda_object, plotting_alpha = 0.25)

plot_gelman_beta(coda_object, env_file = ".env", plotting_alpha = 0.25)

plot_gelman_omega(coda_object, n_omega = 1000L, plotting_alpha = 0.25)

plot_gelman_rho(coda_object)

Arguments

path_coda

Character or mcmc.list. Path to a file containing the coda object, or an mcmc.list object representing MCMC samples.

alpha, beta, omega, rho

Logical. If TRUE, plots the Gelman-Rubin statistic for the respective model parameters (alpha, beta, omega, or rho). Default: TRUE for all parameters.

n_omega

Integer. Number of species sampled for the omega parameter. Default: 1000L.

plotting_alpha

Numeric. Transparency level (alpha) for plot lines (0 = fully transparent, 1 = fully opaque). Default: 0.25.

env_file

Character. Path to the environment file containing paths to data sources. Defaults to .env.

return_plots

Character. Path to the folder where the output plots will be saved.

coda_object

mcmc.list. An MCMC sample object containing posterior distributions from an Hmsc model.

Author

Ahmed El-Gabbas