Bootstrap confidence intervals for each growth model's parameters — ensemble_bootstrap

Resamples the ensemble fit (see ensemble_bootstrap_ci() for the choice between "poisson" and "residual" resampling) and, instead of (or in addition to) the ensemble curve, collects each of the three growth models' own parameter estimates (\(Y_0\), \(\gamma\), \(K\)) and weight on every replicate, returning percentile confidence limits for each. This is what the manuscript's per-segment parameter tables report.

Usage

ensemble_bootstrap_params(
  fit,
  method = c("poisson", "residual"),
  B = 200L,
  level = 0.95,
  beta_fixed = 1.5
)

Arguments

fit: An ensemble_fit object.
method: "poisson" (default) or "residual".
B: Number of bootstrap replicates.
level: Confidence level.
beta_fixed: Passed to fit_ensemble() for each replicate.

Value

A data frame with one row per model x parameter, giving the point estimate, lower/upper confidence limits, and number of successful replicates (B minus failed refits).