confinterval

A lightweight Python toolkit for computing confidence intervals for means, medians, percentiles, and proportions.

It supports:

• t confidence intervals for the mean
• normal approximation confidence intervals for the mean
• bootstrap confidence intervals for mean, median, standard deviation, IQR, and percentiles
• Wilson confidence intervals for proportions

Installation

Intsall confinterval from PyPI

pip install confinterval

Quick Start

from confinterval import ci

data = [18, 19, 20, 21, 22, 20, 19, 21, 20, 22]

result = ci(data, confidence=0.95).t()

print(result)

Output:

mean: 20.2
95% t confidence interval: [19.384, 21.016]

You can also access the values directly:

result.estimate
result.lower
result.upper
result.confidence
result.method
result.statistic

Bootstrap intervals

Bootstrap intervals are useful for statistics like the median or percentiles.

from confinterval import ci

data = [18, 19, 20, 21, 22, 20, 19, 21, 20, 22]

result = ci(data, confidence=0.95, random_state=42).bootstrap("median")

print(result)

Output:

median: 20
95% bootstrap confidence interval: [19, 21]

You can use:

ci(data).bootstrap("mean")
ci(data).bootstrap("median")
ci(data).bootstrap("std")
ci(data).bootstrap("iqr")
ci(data).bootstrap("p90")
ci(data).bootstrap("p95")

Proportion intervals

Use Wilson intervals for proportions, rates, or percentages.

from confinterval import ci

result = ci.proportion(87, 100, confidence=0.95).wilson()

print(result)

This means 87 successes out of 100 total trials.

Output:

proportion: 0.87
95% wilson confidence interval: [0.790196, 0.922428]

Confidence level

The default confidence level is 95%. You can change it:

ci(data, confidence=0.90).t()
ci(data, confidence=0.99).bootstrap("median")
ci.proportion(87, 100, confidence=0.95).wilson()

Higher confidence gives a wider interval.

Convert result to dictionary

data = [18, 19, 20, 21, 22, 20, 19, 21, 20, 22]

result = ci(data).t()

result.to_dict()

Output:

{'estimate': 20.2,
 'lower': 19.25818886972757,
 'upper': 21.14181113027243,
 'confidence': 0.95,
 'method': 't',
 'statistic': 'mean'}

License

This project is under the MIT License.