Non-parametric bootstrap resampling for confidence intervals and standard errors —
mirrors scipy.stats.bootstrap with percentile, basic,
and BCa methods.
Use bootstrap1 for single-sample statistics. The default method is BCa (bias-corrected and accelerated).
All three methods are available via the method option. BCa is generally preferred for skewed distributions.
Any statistic function can be passed — here we compute a 95% BCa confidence interval for the median.
Use bootstrap for multi-sample statistics. Each sample is resampled independently.
The bootDistribution property gives access to all resampled statistic values for custom analysis.