Decompose Prediction Uncertainty Descriptively — prediction_uncertainty

Separates variability in conditional expected-response draws from total posterior predictive variability. The difference is a descriptive Monte Carlo decomposition and is not a causal variance decomposition.

Usage

prediction_uncertainty_decomposition(
  fit,
  newdata = NULL,
  include_group_effects = FALSE,
  allow_new_levels = FALSE,
  ndraws = 1000L,
  seed = 1L
)

Arguments

fit: A fitted gp3bayes_fit.
newdata: Optional data frame. NULL uses the fitted prepared data.
include_group_effects: Whether fitted group-level effects are included.
allow_new_levels: Whether new grouping levels are permitted by brms.
ndraws: Optional number of posterior draws.
seed: Non-negative seed used for posterior predictive simulation.

Value

A gp3bayes_prediction_uncertainty object.