Skip to contents

Plot Calibration Curves (AnPIT / PIT) from Spatial Cross-Validation

plot_AnPIT.Rd

Produce calibration plots from a RiskMap.spatial.cv object returned by assess_pp. * For Binomial or Poisson models the function visualises the Aggregated normalised Probability Integral Transform (AnPIT) curves stored in $AnPIT. * For Gaussian models it detects the list $PIT and instead plots the empirical Probability Integral Transform curve (ECDF of PIT values) on the same \(u\)-grid.

A 45° dashed red line indicates perfect calibration.

Usage

plot_AnPIT(
  object,
  mode = "average",
  test_set = NULL,
  model_name = NULL,
  combine_panels = FALSE
)

Arguments

object

A RiskMap.spatial.cv object.

mode

One of "average" (average curve across test sets), "single" (a specific test set), or "all" (every test set separately).

test_set

Integer; required when mode = "single".

model_name

Optional character string; if supplied, only that model is plotted.

combine_panels

Logical; when mode = "average", draw all models in a single panel (TRUE) or one panel per model (FALSE, default).

Value

A ggplot2 object (single plot) or a ggpubr grid.