Skip to contents

Plot posterior draws

Source: R/posterior.R
plot.RprobitB_fit.Rd

Creates a standard posterior diagnostic or uncertainty plot with bayesplot, or draws the estimated mixing distributions of the random effects.

Usage

# S3 method for class 'RprobitB_fit'
plot(
  x,
  y = NULL,
  type = c("trace", "rank", "acf", "density", "interval", "pairs", "mixture"),
  variables = NULL,
  ...
)

Arguments

x

[RprobitB_fit]
Fitted choice model.

y

[NULL]
Currently not used.

type

[character(1)]
The plot to create:

  • "trace" draws the sampled values of each chain over the iterations.

  • "rank" compares the chains through the ranks of their draws.

  • "acf" draws the autocorrelation within each chain.

  • "density" overlays the marginal posterior density of each chain.

  • "interval" draws posterior point estimates with credible intervals.

  • "pairs" draws bivariate scatter plots of the variables.

  • "mixture" draws the estimated mixing distribution of every random effect, the population density of its coefficient averaged over the retained draws. For an effect with latent classes, it is the mixture of the class densities. For simulated data, a dashed line adds the mixing distribution that generated them.

variables

[character() | NULL]
Posterior variables to include. NULL includes all varying model parameters and excludes individual coefficients and latent allocations. For type = "mixture", the names of random effects as in mu[<effect>], and NULL includes all random effects.

...

Further arguments passed to the selected bayesplot function. Not used for type = "mixture".

Value

A ggplot object or, for a pairs plot, a bayesplot_grid object.

Examples

set.seed(1)
model <- fit(
  choice ~ x + z | 0, dgp_parameters = list(beta = c(x = 1, z = -0.5)),
  chains = 2
)

### convergence and mixing of the chains
plot(model, type = "trace")

plot(model, type = "rank")

plot(model, type = "acf")


### marginal and joint posterior distributions
plot(model, type = "density")

plot(model, type = "interval")

plot(model, type = "pairs")


### estimated mixing distribution of a random coefficient
mixed <- fit(
  choice ~ x | 0, random_effects = "x",
  dgp_parameters = list(beta = c(x = 1), Omega = matrix(0.5)),
  chains = 1, n_occasions = 10
)
plot(mixed, type = "mixture")