reRandomStats Documentation

Tests Docs License: MIT

A comprehensive Python toolkit for re-randomisation statistics in the tradition of Sir Ronald A. Fisher, extended in v0.2.0 with time-stratified case-crossover estimators, model-comparison helpers that share a single algorithmic source for multiple-comparisons correction, and a dose-response / breakpoint-analysis toolkit (broken-stick + Davies + Muggeo Pseudo-Score + 4-parameter Hill).

Features

Core (v0.1.0+):

  • Fisher’s Resampling Test — flexible test statistics (mean / median / sum differences)

  • Fisher’s Exact Test — 2×2 contingency table

  • Multi-Group Pairwise Testing — with automatic multiple-testing correction

  • Binomial Proportion Tests — single- and two-sample variants with Wilson confidence intervals

  • Classical Hypothesis Tests — unified dispatcher for Mann-Whitney U, Kruskal-Wallis, Kolmogorov-Smirnov, Mood’s Median, Wilcoxon Rank-Sum, independent t-test, chi-square

  • Data I/O — CSV reader with German-locale support and wide → long table conversion

  • Pretty-Table Output — manuscript-ready ASCII / Markdown tables

New in v0.2.0 — three submodules:

  • case_crossover — time-stratified case-crossover conditional logit (Maclure 1991, Lee et al. 2023) with stratified-permutation backup, daylight-hours covariate, within-event temporal-contrast test (hot-day-vs-hot-week), and Burke-2015 σ-rescaled effect translator

  • model_comparison — two-sample Wald z-test on independently-estimated coefficients, nested-model likelihood- ratio test, single-method correction (correct_pvalues), array helper (correct_pvalues_array), and dual-method BH + Bonferroni report (benjamini_hochberg); all four correction helpers route through one statsmodels.stats.multitest.multipletests call

  • dose_response — broken-stick segmented regression with profile-RSS 95 % CI on the breakpoint, Davies (1987 / 2002) and Muggeo (2016) Pseudo-Score tests for breakpoint existence, 4-parameter Hill fit with Sebaugh–McCray (2003) lower-bend point, and a per-subject iterator that applies any of the four to a panel of subjects (pickle-safe for concurrent.futures.ProcessPoolExecutor)

Quick Start

Core resampling test:

from rerandomstats import FisherResamplingTest

test = FisherResamplingTest(
    data_a=[1.2, 3.4, 2.1, 4.5, 3.3],
    data_b=[5.6, 7.8, 6.5, 8.9, 7.2],
    func='medianDiff',
    combination_n=20_000,
)
p_value = test.main()
print(f"p = {p_value:.4f}")

Multi-comparisons correction on a heterogeneous battery:

from rerandomstats import benjamini_hochberg

# p-values from arbitrary tests (case-crossover, Wald, Poisson, LRT, …)
result = benjamini_hochberg({
    "H1": 0.001, "H2": 0.012, "H3": 0.040, "H4": 0.080,
})
for name, row in result["results"].items():
    print(f"{name}: q = {row['bh_adjusted_p']:.4f}  "
          f"BH-reject = {row['bh_reject']}")

Breakpoint detection on dose-response data:

import numpy as np
from rerandomstats import broken_stick_fit, davies_test, hill_fit

x = np.random.uniform(10, 30, 300)
y = np.where(x <= 22, 38 + 0.05 * x,
             38 + 0.05 * 22 + 0.8 * (x - 22))
y += np.random.normal(0, 0.2, 300)

bs = broken_stick_fit(x, y)
print(f"Breakpoint = {bs['breakpoint']:.2f} "
      f"[{bs['breakpoint_ci_lo']:.2f}, {bs['breakpoint_ci_hi']:.2f}]")

davies = davies_test(x, y)
print(f"Davies p = {davies['pvalue']:.4f}")

hill = hill_fit(x, y)
print(f"EC50 = {hill['ec50']:.2f}, Hill n = {hill['hill_n']:.2f}, "
      f"lower bend = {hill['lower_bend']:.2f}")

Contents

Indices and tables