Pingouin
Pingouin is a free, open-source statistical package written in Python 3 and built mostly on Pandas and NumPy.
What Pingouin is
Pingouin is a free, open-source statistical package written in Python 3 and built mostly on Pandas and NumPy. It provides statistical tests, regression models, Bayesian analysis, and scripting support, with CSV and XLSX imports through Pandas. Correlation tools cover robust, partial, distance, repeated-measures, and pairwise methods. Modeling functions include linear and logistic regression, mediation analysis, and multivariate tests. The package calculates effect sizes, performs power analysis, and produces parametric or bootstrapped confidence intervals for effect sizes and correlation coefficients. Plotting functions include Bland–Altman, Q-Q, and paired plots. Several features can run as Pandas DataFrame methods, including ANOVA, pairwise tests, correlations, and mediation analysis. Installation is available through uv, pip, or conda. Documentation lists Python 3.11 and later testing and dependencies such as SciPy, Statsmodels, Matplotlib, Seaborn, and Scikit-learn. The project is under development, warns that bugs may occur, recommends checking results with other statistical software, and provides no warranty. It is licensed GPL-3.0 and maintained by Raphael Vallat.
Good
- Free and open-source under GPL-3.0
- Broad correlation and regression methods
- Effect sizes and power analysis included
- Pandas DataFrame method support
- CSV and XLSX import through Pandas
Watch out
- Project is still under development
- Documentation warns bugs may exist
- No warranty is provided
- Requires Python and listed dependencies
Who it suits
Pingouin suits Python users who need open-source statistical analysis integrated with Pandas. It is appropriate for scripting, regression, correlations, Bayesian work, and power analysis.
Verdict
Pingouin offers a wide statistical toolkit inside Python and Pandas workflows, including modeling, correlations, confidence intervals, and plots. Its development status means results should be checked with other statistical software.
Plans and prices
From Pingouin's own pricing page, checked 12 Oct 2001.
| Plan | Price | What it gives you |
|---|---|---|
| Free and open-source | Free | Python package; no paid plans or usage limits stated |
The facts
Each one read from Pingouin's own pages, with the date it was checked.
What it does
- Bayesian analysis
- Pingouin includes Bayes factors.checked 12 Oct 2001
- Correlations
- It supports robust, partial, distance, repeated-measures, and pairwise correlations.checked 12 Oct 2001
- Data import formats
- .csv, .xlsx via Pandaschecked 23 Sep 2001
- Development status
- The project says it is under development, warns that bugs may exist, and advises checking results with other statistical software; it provides no warranty.checked 12 Oct 2001
- Effect sizes and power
- It provides effect sizes, power analysis, and parametric or bootstrapped confidence intervals for effect sizes or correlation coefficients.checked 12 Oct 2001
- Installation
- The package can be installed with uv, pip, or conda.checked 12 Oct 2001
- License
- The project’s GitHub repository identifies its license as GPL-3.0.checked 12 Oct 2001
- Maintainer
- The documentation says Pingouin was created and is maintained by Raphael Vallat, mostly in his spare time.checked 12 Oct 2001
- Modeling
- Its listed functions include linear and logistic regression, mediation analysis, and multivariate tests.checked 12 Oct 2001
- Pandas integration
- Several functions can be called directly as pandas DataFrame methods, including ANOVA, pairwise tests, correlations, and mediation analysis.checked 12 Oct 2001
- Purpose
- Pingouin is an open-source statistical package written in Python 3 and based mostly on Pandas and NumPy.checked 12 Oct 2001
- Regression models
- Yeschecked 23 Sep 2001
- Runtime requirements
- The documentation says Pingouin is tested for Python 3.11 and later and lists NumPy, SciPy, Pandas, Statsmodels, Matplotlib, Seaborn, Scikit-learn, and other dependencies.checked 12 Oct 2001
- Scripting support
- Yeschecked 23 Sep 2001
- Statistical tests
- It includes N-way, repeated-measures, mixed ANOVA and ANCOVA, pairwise tests, and chi-squared tests.checked 12 Oct 2001
- Support
- The documentation directs questions to GitHub Discussions and bug reports to the GitHub repository.checked 12 Oct 2001
- Visualization
- Its plotting functions include Bland–Altman, Q-Q, and paired plots.checked 12 Oct 2001
