Recompute SBC statistics without refitting models. — recompute_SBC

Useful for example to recompute SBC ranks with a different choice of thin_ranks or added derived quantities.

recompute_SBC_statistics(
  old_results,
  datasets,
  backend,
  thin_ranks = SBC_backend_default_thin_ranks(backend),
  ensure_num_ranks_divisor = 2,
  dquants = NULL,
  gen_quants = NULL
)

Arguments

datasets: an object of class SBC_datasets
backend: backend used to fit the results. Used to pull various defaults and other setting influencing the computation of statistics.
thin_ranks: how much thinning should be applied to posterior draws before computing ranks for SBC. Should be large enough to avoid any noticeable autocorrelation of the thinned draws See details below.
ensure_num_ranks_divisor: Potentially drop some posterior samples to ensure that this number divides the total number of SBC ranks (see Details).
dquants: Derived quantities to include in SBC. Use derived_quantities() to construct them.
gen_quants: Deprecated, use dquants instead

Value

An S3 object of class SBC_results with updated $stats and $default_diagnostics fields.