bootstrapx

bootstrapx is a Python library for production-grade bootstrap uncertainty estimation.

It is designed for cases where scipy.stats.bootstrap is not enough: time-series dependence, cluster-aware resampling, weighted bootstrap, studentized intervals, and production-safe reproducibility.

Why bootstrapx

• 15+ bootstrap methods in one API.
• Reproducible random-state handling.
• Memory-safe batched computation.
• Integrations for pandas and scikit-learn.
• Focus on real-world analytics, ML evaluation, and time-series work.

Where it fits

Use bootstrapx when you need one of these:

• BCa or studentized intervals for a custom statistic.
• Block bootstrap for dependent time series.
• Cluster or stratified bootstrap for grouped data.
• Bootstrap uncertainty around model metrics.
• A consistent API that works from notebooks to backend services.

Quick example

import numpy as np
from bootstrapx import bootstrap

data = np.random.default_rng(42).normal(5, 2, size=300)
result = bootstrap(data, np.mean, method="bca", n_resamples=4999, random_state=42)

print(result)
print(result.confidence_interval.low, result.confidence_interval.high)

Main sections

• Getting Started: install and first examples.
• Methods: which bootstrap method to choose and why.
• Integrations: pandas and scikit-learn workflows.
• A/B Testing: cluster-aware patterns and practical guidance.
• Time Series: block, stationary, sieve, and wild bootstrap.
• API Reference: public interfaces and examples.