Parametric bootstrap confidence bands for a distribution — bootstrap

Fit distribution to data, then resample it replications times to derive percentile confidence bands on the quantile curve. Wraps the shared C++ BootstrapAnalysis (Numerics).

Usage

bootstrap_analysis(
  data,
  distribution,
  probabilities,
  estimation_method = "MaximumLikelihood",
  sample_size = NULL,
  replications = 1000L,
  seed = 12345L,
  alpha = 0.1
)

Arguments

data: numeric vector of observations.
distribution: distribution family name to fit + bootstrap.
probabilities: numeric vector of non-exceedance probabilities at which to tabulate the quantile curve + confidence band.
estimation_method: fit method: "MaximumLikelihood" (default), "MethodOfMoments", or "MethodOfLinearMoments".
sample_size: bootstrap sample size (default NULL, uses length(data)).
replications: number of bootstrap replications (default 1000).
seed: PRNG seed for the bootstrap.
alpha: significance level for the two-sided percentile band (default 0.1 -> 90% band).

Value

A named list: parameters (fitted), mode_curve (fitted-distribution quantiles), mean_curve, lower_ci, upper_ci (percentile band, one entry per probability).