Run statistical simulations and analysis

Implement bootstrap, permutation, and Monte Carlo statistical procedures

Deriving confidence intervals and p-values when analytical formulas are unavailable or standard assumptions violated.

• · raw data
• · statistic to estimate (mean/median/correlation/etc)
• · resampling count (B=1000-10000)

• · bootstrap distribution
• · confidence intervals (percentile/BCa/Studentized)
• · p-values from permutation tests

• · numpy/scipy PRNG
• · statistical computation libraries

Data should be i.i.d. for nonparametric bootstrap; samples >30 preferred; infinite variance distributions invalid

• · Extreme quantiles unreliable (tails)
• · Dependent data without block bootstrap
• · Very small samples (<10) insufficient resampling
• · Non-i.i.d. data violates bootstrap assumption

• · Efron bootstrap lineage cited
• · BCa bias-corrected intervals documented
• · Block bootstrap for time series mentioned