Skip to contents

Plot MCMC Chains for Drift Diffusion Model Parameters

Source: R/plotting.R
plot_mcmc_trace.Rd

The functions provide visualizations of MCMC results obtained for drift_dm() objects. Users won't call the functions directly. Instead, they are called via plot.mcmc_dm(), and the following documentation helps to define optional arguments that can be passed on via ....

Usage

plot_mcmc_trace(
  chains,
  col_palette = grDevices::rainbow,
  col_chains = NULL,
  which_prms = NULL,
  xlab = NULL,
  ylab = NULL,
  xlim = NULL,
  ylim = NULL
)

plot_mcmc_marginal(
  chains,
  col_line = NULL,
  col_shade = NULL,
  which_prms = NULL,
  xlab = NULL,
  ylab = NULL,
  xlim = NULL,
  ylim = NULL
)

plot_mcmc_auto(
  chains,
  lags = 1:30,
  col_line = NULL,
  which_prms = NULL,
  xlab = NULL,
  ylab = NULL,
  xlim = NULL,
  ylim = NULL,
  type = NULL,
  main = NULL
)

Arguments

chains

an array of MCMC samples with three dimensions (parameters × chains × iterations; see also plot.mcmc_dm()). This argument is not optional and will be provided by plot.mcmc_dm().

col_palette

a function to generate a color palette for chains (default is grDevices::rainbow). Must be callable like this col_palette(n).

col_chains

a character vector, defining colors(s) to override col_palette for chain lines. Can be a single value or a vector matching the number of chains. Recycled if necessary to match the number of chains.

which_prms

a regular expression (string), used to select a subset of parameters to plot. For example, "^v" would match all parameters starting with v. See also the examples in plot.mcmc_dm().

xlab, ylab

character vector(s), specifying axis labels for each panel. Recycled if necessary to match the number of plots being generated.

xlim, ylim

list(s), containing length-2 numeric vectors specifying x- and y-axis limits, respectively. Recycled if necessary to match the number of plots being generated.

col_line

a character vector, defining color(s) for the line outlines in the density plots. Can be a single value or a vector matching the number of parameters to plot. Recycled if necessary to match the number of parameters being plotted.

col_shade

a character vector, defining color(s) for the shaded areas in the density plots. Can be a single value or a vector matching the number of parameters to plot. Recycled if necessary to match the number of parameters being plotted.

lags

a numeric vector, giving the lags at which to calculate autocorrelation (see also coda::autocorr.diag() and coda::autocorr()).

type

a character, specifying how autocorrelations shall be displayed. Defaults to "h" (see the type argument of base::plot()). Recycled if necessary to match the number of parameters being plotted.

main

character vector, specifying the labels above each autocorrelation plot. Defaults to parameter labels. Recycled if necessary to match the number of parameters being plotted.

Value

These functions are called for their side effects (producing plots). They return NULL invisibly.

Details

  • plot_mcmc_trace() plots one panel per parameter, with lines for each chain showing how values evolve over iterations. This is useful for diagnosing convergence and mixing.

  • plot_mcmc_marginal() plots smoothed marginal posterior densities collapsed over chains and iterations for each parameter, useful for inspecting posterior distributions.

  • plot_mcmc_auto() plots the autocorrelation among samples as function of lags.

See also

plot.mcmc_dm()