Skip to contents

Diagnose Bayesian MCMC fit

Source: R/model_diagnostics.R
diagnose_fit.Rd

Extracts effective sample size (n_eff) and Rhat diagnostics from a fitted Bayesian model (e.g., Stan fit) and provides categorical interpretations along with practical recommendations for improvement.

Usage

diagnose_fit(
  fit,
  n_eff_threshold_high = 1000,
  n_eff_threshold_ok = 500,
  rhat_excellent = 1.01,
  rhat_ok = 1.05,
  include_lp = TRUE
)

Arguments

fit

A fitted Bayesian model object (e.g., stanfit).

n_eff_threshold_high

Numeric. Threshold above which n_eff is considered "high" (default = 1000).

n_eff_threshold_ok

Numeric. Threshold above which n_eff is considered "moderate" (default = 500).

rhat_excellent

Numeric. Threshold below which Rhat is considered "excellent" (default = 1.01).

rhat_ok

Numeric. Threshold below which Rhat is considered "acceptable" (default = 1.05).

include_lp

Logical. If FALSE, excludes lp from the output table (default = TRUE).

Value

A data frame with columns

  • n_eff effective sample size

  • Rhat Gelman-Rubin convergence diagnostic

  • Rhat_interpretation categorical interpretation of Rhat

  • n_eff_interpretation categorical interpretation of n_eff

  • recommendation practical suggestions if diagnostics are poor

Examples

if (FALSE) {
diagnose_fit(bayesian_fit_kn199$stan_fit)
diagnose_fit(bayesian_fit_kn199$stan_fit, include_lp = FALSE)
}