Skip to contents

Compare observed data with replicated outcomes from an estimated BayesRTMB model. Density overlays are used for continuous outcomes and binned bar plots for discrete outcomes. When `type = "auto"`, the display is selected from the likelihood's `_lpdf` or `_lpmf` implementation.

Usage

pp_check(object, ...)

# S3 method for class 'RTMB_Fit_Base'
pp_check(object, ...)

# S3 method for class 'rtmb_pp_check'
plot(
  x,
  main = NULL,
  xlab = NULL,
  ylab = NULL,
  observed_col = "#1B1B1B",
  predictive_col = "#2C7FB8",
  show_legend = TRUE,
  legend_position = c("auto", "topleft", "topright", "bottomleft", "bottomright"),
  legend_cex = 0.78,
  interval = 0.95,
  ...
)

Arguments

object

A BayesRTMB fit object.

...

Arguments passed to the fit object's `pp_check()` method or to the plotting method.

x

An `rtmb_pp_check` object.

main

Optional plot title.

xlab

Optional x-axis label.

ylab

Optional y-axis label.

observed_col

Color for observed data.

predictive_col

Color for predictive data and intervals.

show_legend

Logical; display the plot legend.

legend_position

Legend position. `"auto"` moves scatter-plot legends away from the fitted trend; the other values are passed to `legend()`.

legend_cex

Relative text and symbol size for the legend.

interval

Probability covered by posterior predictive intervals.

Value

An object of class `rtmb_pp_check`, returned invisibly after plotting.

Details

Supplying `x` to the fit object's `pp_check()` method switches to a scatter-based check. Observed outcomes are compared with posterior predictive means and 95 value `x = ".fitted"` uses the posterior predictive mean on the horizontal axis for a model-wide calibration check.