hmer (version 1.6.2)

wave_values: Multiple Wave Output Plotting

Description

Given multiple waves of points, produces pairs plots of the outputs.

Usage

wave_values(
  waves,
  targets,
  output_names = names(targets),
  ems = NULL,
  surround = FALSE,
  restrict = FALSE,
  p_size = 1.5,
  l_wid = 1.5,
  zero_in = TRUE,
  wave_numbers = ifelse(zero_in, 0, 1):(length(waves) - ifelse(zero_in, 0, 1)),
  which_wave = ifelse(zero_in, 0, 1),
  upper_scale = 1,
  ...
)

Value

A ggplot object.

Arguments

waves

The list of data.frames, one for each set of outputs at that wave.

targets

The output targets.

output_names

The outputs to plot.

ems

If provided, plots the emulator expectations and 3-standard deviations.

surround

As in wave_points.

restrict

Should the plotting automatically restrict to failing target windows?

p_size

As in wave_points.

l_wid

The width of the lines that create the target boxes.

zero_in

Is a wave 0 included in the waves list?

wave_numbers

Which waves to plot.

which_wave

Scaling for lower plots (see description)

upper_scale

Scaling for upper plots (ibid)

...

Optional parameters (not to be used directly)

Details

This function operates in a similar fashion to wave_points - the main difference is that the output values are plotted. Consequently, the set of targets is required to overlay the region of interest onto the plot.

To ensure that the wave numbers provided in the legend match, one should provide waves as a list of data.frames with the earliest wave at the start of the list.

The parameters which_wave and upper_scale control the level of `zoom' on each of the lower-triangular and upper-triangular plots, respectively. For the lower plots, which_wave determines which of the provided waves is to be used to determine the output ranges to plot with respect to: generally, higher which_wave values result in a more zoomed-in plot. For the upper plots, upper_scale determines the plot window via a multiple of the target bounds: higher values result in a more zoomed-out plot. If not provided, these default to which_wave=0 (or 1 if no wave 0 is given) and upper_scale = 1. If the value provided to which_wave does not correspond to a provided wave (or one explicitly not included in wave_numbers), it defaults to the closest available wave to the value of which_wave.

If ems is provided, it should follow the same structure as waves: at the very least, it should contain all emulators trained over the course of the waves. The emulator predictions for a target are made by the emulator for that target whose ranges are the smallest such that contain the point.

See Also

Other visualisation tools: behaviour_plot(), diagnostic_wrap(), effect_strength(), emulator_plot(), hit_by_wave(), output_plot(), plot_actives(), plot_lattice(), plot_wrap(), simulator_plot(), space_removed(), validation_pairs(), wave_dependencies(), wave_points()

Examples

 wave_values(SIRMultiWaveData, SIREmulators$targets, surround = TRUE, p_size = 1)
 # \donttest{
   wave_values(SIRMultiWaveData, SIREmulators$targets, c('nS', 'nI'), l_wid = 0.8)
     wave_values(SIRMultiWaveData, SIREmulators$targets, l_wid = 0.8,
      wave_numbers = c(0, 1, 3), which_wave = 2, upper_scale =  1.5)
     # For many plots, it may be helpful to manually modify the font size
     wave_values(SIRMultiWaveData, SIREmulators$targets) +
      ggplot2::theme(text = ggplot2::element_text(size = 5))
 # }