reliability

What it is and what it does

Reliability is a Python library for reliability engineering and survival analysis that extends scipy.stats with specialized tools for fitting probability distributions, modeling failure mechanisms, and analyzing time-to-failure data. It supports Weibull, Exponential, Gamma, Gumbel, Normal, Lognormal, Loglogistic, and Beta distributions, with capabilities for handling right-censored data, mixture models, and competing risks. The library includes non-parametric survival estimation (Kaplan-Meier, Nelson-Aalen), goodness-of-fit testing, and interactive visualization via matplotlib.

The package is designed for engineers and researchers working with reliability data who need tools otherwise available only in proprietary software. It covers accelerated life testing with multiple life-stress models, physics-of-failure analysis, repairable systems modeling, and reliability growth analysis. Dependencies on autograd, scipy, numpy, matplotlib, and pandas provide the numerical and visualization backbone.

Use it for:

• Fit Weibull or other distributions to product failure data and estimate reliability metrics like mean time to failure.
• Analyze survival curves from clinical or industrial time-to-event studies with censored observations using Kaplan-Meier estimation.
• Model accelerated life testing experiments with temperature, voltage, or stress as covariates across multiple distributions.
• Calculate stress-strength interference probabilities to assess design margin between applied stress and material strength.
• Generate probability plots and goodness-of-fit statistics to compare competing distribution models for a dataset.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Reliability engineering and survival analysis library that fits probability distributions to data, models failure mechanisms, and provides tools for accelerated life testing and repairable systems analysis.

Yes, if you work in reliability engineering, quality assurance, or survival analysis and need distribution fitting and specialized statistical tools. The library is production-stable with no known vulnerabilities and low install friction. The LGPLv3 copyleft license requires derivative works to remain open-source. Maintenance is aging (525 days since last release), so verify that the feature set and numerical methods meet your current needs before committing to a long-term dependency.

Install

reliability on PyPI

pip install reliability

uv add reliability

poetry add reliability

Installing reliability

Before you install

Low install friction with a pure-wheel distribution. Maintenance is aging—last release was 525 days ago—but the repository remains active with recent commits and no archival.

License in practice

Licensed under LGPLv3 (copyleft). Derivative works and modifications must be distributed under the same license; proprietary closed-source use is not permitted.

Quickstart

pip install reliability

from reliability.Distributions import Weibull_Distribution
dist = Weibull_Distribution(alpha=4, beta=2)
print(dist.mean)

Requires Python 3.10 or later.

Verify before relying

• Whether the 7 runtime dependencies (autograd, scipy, numpy, matplotlib, pandas, autograd-gamma, mplcursors) are all required for basic use or only for specific features.
• Current performance and numerical stability of the 24 accelerated life testing models on large datasets.

Package facts

Evidence: reliability-0.9.0-py3-none-any.whl

Keywords: reliability, engineering, RAM, weibull, lognormal, exponential, beta, gamma, normal, loglogistic, gumbel, extreme, value, kaplan meier, kaplan-meier, survival, analysis, censored, data, lifelines, probability, distribution, distributions, fit, fitting, curve, quality, ALT, accelerated, life, testing, MCF, mean, cumulative, CIF, DS, ZI, defective, subpopulation, zero, inflated, DSZI, likelihood

Tags

More Scientific/Engineering packages