Plot the DEoptim search trace

Draws the best objective value reached by each generation. When stats is supplied, the loss it records is added as a point at the last generation: the search reports the minimum of many noisy evaluations and so is optimistic about its own winner, and re-simulating that winner at a higher replicate count is the honest measurement. The gap between line and point is that optimism.

Usage

plot_calibration_convergence(trace, stats = NULL)

Arguments

trace

The DEoptim trace: either a path to the CSV calibrate_dynamic_fire() writes (columns generation and best value) or a data frame with those two columns in that order.

stats

Optional run_calibration_validation() summary, used for the re-checked loss point and the replicate count in its label.

Value

A ggplot object.

See also

Other calibration plots: calibration_events(), calibration_plot_palette(), plot_calibration_area_by_fuel(), plot_calibration_fire_counts(), plot_calibration_fire_sizes(), plot_calibration_loss(), plot_calibration_severity(), plot_calibration_severity_by_size()