validation_diagnostics: Emulator Diagnostics

In Tandethsquire/emulatorr: Emulation and History Matching Package

Description

Plots standard diagnostics for emulators.

Usage

validation_diagnostics(
  emulators,
  validation_points,
  output_names,
  which_diag = "all",
  targets = NULL,
  ...
)

Arguments

emulators	A list of emulators on which to perform diagnostics
validation_points	The validation set
output_names	The list of outputs to perform diagnostics on
which_diag	Which diagnostics should be performed?
targets	If required, the list of observations for the outputs
...	Any additional parameters to pass to the diagnostic tests.

Details

These diagnostics are based on having two datasets: a training set and a validation set. The emulators will have been trained on the training set, and the validation set is passed to the functions in this wrapper.

The current options for diagnostics (with the codes for which_diag) are:

Standard Errors (se)

Comparison Diagnostics (cd)

Classification Error (ce)

All of the above (all)

For details on each of these, see the help files for standard_errors, comparison_diagnostics and classification_error respectively.

Value

A data.frame containing the points that failed one or more diagnostic tests.

Examples

output_names <- c('nS','nI','nR')
ems <- emulator_from_data(GillespieSIR, output_names,
 ranges = list(aSI = c(0.1, 0.8), aIR = c(0, 0.5), aSR = c(0, 0.05)),
 quadratic = TRUE)
targets <- list(
 list(val = 281, sigma = 10.43),
 list(val = 30, sigma = 11.16),
 list(val = 689, sigma = 14.32)
)
validation_diagnostics(ems, GillespieValidation, output_names, targets = targets)
validation_diagnostics(ems, GillespieValidation, output_names, c('se','cd'))
validation_diagnostics(ems[1:2], GillespieValidation, output_names[1:2], 'ce', targets[1:2])
validation_diagnostics(ems, GillespieValidation, output_names,
 targets = targets, sd = 1, cutoff = 4)

Tandethsquire/emulatorr documentation built on April 12, 2021, 1:08 a.m.