Skip to contents

Plot posterior predictive checks for a fitted probit model

Source: R/probit_posterior_predictive.R
plot_probit_posterior_predictive_checks.Rd

Compact multi-page PDF: observed-vs-replicated marginal resistance, number-resistant-per-event distribution, observed-vs-replicated pairwise RR/RS/SR/SS, complete-profile concentration/entropy, hospital heterogeneity, admission/patient clustering, and (when ppc_replicates with materialised replicate arrays is supplied) a small-multiples page comparing the observed data against n_small_multiples randomly selected posterior replications. A good posterior predictive plot makes it visually apparent whether the observed data look typical among the replicated datasets.

Usage

plot_probit_posterior_predictive_checks(
  ppc_statistics,
  ppc_replicates = NULL,
  output_pdf_path,
  title_base,
  n_small_multiples = 10L,
  small_multiple_seed = 123L
)

Arguments

ppc_statistics

Tibble returned by compute_probit_ppc_statistics.

ppc_replicates

Optional "amr_ppc_draws" object returned by simulate_probit_posterior_predictive with return_replicates = TRUE – enables the small-multiples page. NULL (default) skips that page.

output_pdf_path

Path to the PDF to write.

title_base

Character; prefixed to every page's title (e.g. "<experiment_id>\n<pathogen>").

n_small_multiples

Integer; number of replicate states to show on the small-multiples page. Default 10.

small_multiple_seed

Integer seed for selecting which replicate states appear on the small-multiples page. Default 123.

Value

Invisibly returns the path to the saved PDF, or NULL if skipped.