Posterior predictive check — posterior_predictive

Fit a Bayesian MCMC over a model, draw replicate datasets from the posterior, and compute the common posterior predictive p-values (mean / SD / skewness / min / max) plus a misfit flag. Wraps the shared C++ PosteriorPredictiveCheck.

Usage

posterior_predictive_check(
  data,
  distribution,
  sampler = "DEMCz",
  iterations = 3000L,
  output_length = 10000L,
  seed = 12345L,
  number_of_replicates = 1000L,
  thinning_interval = -1L
)

Arguments

data: numeric vector of observations (also the observed data for the p-value comparison).
distribution: distribution family name.
sampler: MCMC sampler: "DEMCz" (default), "DEMCzs", "ARWMH", or "NUTS".
iterations: number of post-warmup MCMC iterations.
output_length: number of posterior samples used to build the credible band.
seed: PRNG seed for the sampler (fixed for reproducibility).
number_of_replicates: number of posterior replicate datasets (default 1000).
thinning_interval: MCMC thinning interval; -1 (default) keeps the sampler's own default.

Value

A named list: number_of_replicates, mean_p_value, sd_p_value, skewness_p_value, min_p_value, max_p_value, and has_misfit (1 when any p-value is extreme, else 0).