mescla.uncertainty.resampling.bootstrap_endmembers

mescla.uncertainty.resampling.bootstrap_endmembers(endmember_samples, samples, sd_samples=None, n_draws=1000, percentiles=(2.5, 97.5), estimator=None, seed=0)

Bootstrap the replicates defining each end-member.

Resamples each end-member’s replicates with replacement, recomputes its mean, and re-runs the mixing calculation. Gives percentile intervals rather than the jackknife’s symmetric ones, which is the better choice when a fraction sits near zero and its distribution is skewed.

Parameters are as jackknife_endmembers(), plus n_draws, percentiles and seed.

Parameters:

• endmember_samples (dict[str, Any])

• samples (Any)

• sd_samples (Any)

• n_draws (int)

• percentiles (tuple[float, float])

• estimator (Any)

• seed (int | None)