Skip to contents

For every species and parameter, shows how the shape error varies as that parameter moves below or above its calibrated value, marginalizing over the others. A parameter whose boxes separate cleanly is one the fit is sensitive to; a parameter whose boxes overlap is one the reference data cannot constrain.

Usage

plot_growth_factorial_sensitivity(scores, current)

Arguments

scores

A tibble from the growth_factorial_scores target.

current

A tibble of the parameters currently in use, with species and the parameter columns.

Value

A ggplot.

Details

Only the SWEPT parameters appear, and which those are is read from scores rather than assumed. biomass_max is pinned across the factorial and recovered arithmetically, so it has no sensitivity to show; a shape parameter the design fixes to one value per species has none either, and drawing it would be actively misleading. Such a parameter has no "calibrated" box at all – every cell sits on one side of the in-use value – so the panel would show a lone box under "lower" or "higher" and invite the reader to interpret the side as a result, when it only restates which value was assigned.

Candidates are shown by their position RELATIVE to the calibrated value rather than by their absolute value: the candidates differ per species, so an absolute axis would have to carry every species' values in every panel.