Summarise prediction simulations into intervals — summarise

Takes the simulated prediction array and computes summary statistics (fit, lower bound, upper bound) based on the specified statistic type and confidence level. Handles probability transformation and the `which = "all"` case with component-wise summaries.

Usage

summarise_predictions(
  yhat_arr,
  level,
  stat.type,
  predict.type,
  N,
  merMod = NULL,
  which.eff = "full",
  pi.comps = NULL,
  is.glmm.with.response = FALSE
)

Arguments

yhat_arr: matrix of simulated predictions with dimensions (N x n.sims), or for `which = "all"`, the combined yhat from `combine_components()`.
level: numeric confidence level (e.g., 0.8 for 80
stat.type: character either "median" or "mean" for the central estimate.
predict.type: character either "linear.prediction" or "probability".
N: integer number of observations in the original newdata.
merMod: merMod object, required when `predict.type = "probability"` to access the link function.
which.eff: character the effect type: "full", "fixed", "random", or "all".
pi.comps: list of component prediction matrices (from `combine_components()` when `which = "all"`), or NULL for other cases.

Value

data.frame with columns `fit`, `upr`, `lwr`. For `which = "all"`, includes additional columns `effect` and `obs`.