Posterior Prediction for Approved gp3bayes Models

Distinguishes conditional expectations, new-outcome posterior predictions, linear-predictor draws, and the conditional median for lognormal duration models.

Usage

predict_model(
  fit,
  newdata = NULL,
  type = c("expected", "predictive", "linear", "median"),
  include_group_effects = FALSE,
  allow_new_levels = FALSE,
  ndraws = NULL,
  probs = c(0.025, 0.5, 0.975),
  seed = 1L
)

Arguments

fit

A fitted gp3bayes_fit.

newdata

Optional data frame. NULL uses the fitted prepared data.

type

Prediction quantity: "expected", "predictive", "linear", or "median".

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.

probs

Three probabilities used to summarise predictions.

seed

Non-negative seed used for posterior predictive simulation.

Value

A gp3bayes_prediction containing draws, summaries, prediction data, and interpretation metadata.