Plot a bootstrap parameter distribution

Description

One histogram per parameter of the bootstrap estimates - the panel PsN generates through xpose4’s boot.hist - with the original estimate drawn as a solid line and the confidence interval as dashed lines. When the run used dofv = TRUE, a final panel shows the distribution of the change in objective function value against its chi-square reference density, on chi_square_df degrees of freedom.

Usage

# S3 method for ferx_bootstrap
plot(x, y = NULL, parameters = NULL, breaks = "FD", ...)

Arguments

  • x: A ferx_bootstrap object.
  • y: Ignored. Present only for consistency with the plot(x, y, ...) generic; must be NULL if supplied.
  • parameters: Character vector selecting which parameters to plot. NULL (the default) plots all of them.
  • breaks: Passed to [hist](hist). Default "FD".
  • ...: Ignored.

Details

The histograms show every replicate that produced an estimate, including any the exclusion filters dropped from the statistics; the lines drawn over them come from $parameters and so reflect the included set only. The counts in $diagnostics say how far the two differ. (Deciding which replicate is excluded is the engine’s job - see [ferx_bootstrap_summarize](ferx_bootstrap_summarize.qmd) to change the criteria.)

Seealso

Other diagnostics: [check_diagnostics](check_diagnostics.qmd), [ferx_conddist](ferx_conddist.qmd), [ferx_cov_screen](ferx_cov_screen.qmd), [ferx_get_warnings](ferx_get_warnings.qmd), [ferx_runlog_iters](ferx_runlog_iters.qmd), [ferx_trace](ferx_trace.qmd), [plot.ferx_fit](plot.ferx_fit.qmd), [summary.ferx_fit](summary.ferx_fit.qmd)

Concept

diagnostics

Value

Invisibly, the character vector of parameters plotted.

Examples

bs <- ferx_bootstrap(ferx_example("warfarin")$model, samples = 200)
plot(bs)
plot(bs, parameters = c("TVCL", "TVV"))