scikit-posthocs

What it is and what it does

scikit-posthocs fills a gap in Python's statistical ecosystem by offering a comprehensive suite of post hoc tests for pairwise multiple comparisons—the follow-up analysis performed after ANOVA or similar omnibus tests reveal statistical significance. It implements both parametric tests (like TukeyHSD and Scheffe) and non-parametric alternatives (like Dunn and Nemenyi tests) for different experimental designs, plus outlier detection methods and basic plotting. The package is tightly integrated with pandas DataFrames and NumPy arrays, making it convenient to work with typical data science workflows.

You would use this package when you've run an ANOVA or Kruskal-Wallis test and need to determine which specific group pairs differ significantly. It handles p-value adjustment automatically to control for multiple comparisons, and it supports both factorial and block designs. The package depends on scipy, statsmodels, pandas, seaborn, and matplotlib, so it fits naturally into existing scientific Python environments.

Use it for:

• Run post hoc pairwise comparisons after ANOVA to identify which group means differ significantly.
• Apply non-parametric post hoc tests (Dunn, Nemenyi) when data violates normality assumptions.
• Detect outliers in datasets using Grubbs, Tietjan-Moore, or ESD tests before analysis.
• Generate significance plots to visualize pairwise comparison results for reports or publications.
• Adjust p-values across multiple comparisons to control Type I error rates in exploratory studies.
• Analyze block design experiments using specialized tests like Durbin-Conover or Quade.

Worth the install?

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

Provides post hoc statistical tests for pairwise multiple comparisons after ANOVA, including parametric tests (Scheffe, Student T, Tamhane T2, TukeyHSD) and non-parametric tests (Dunn, Nemenyi, Mann-Whitney, Wilcoxon) with p-value adjustment and outlier detection.

Yes. The package is actively maintained, has no known vulnerabilities, uses a permissive MIT license, and fills a genuine gap in Python's statistical toolkit by providing post hoc tests that are either missing or inconvenient in scipy and statsmodels. Install friction is low, and it integrates well with standard data science libraries. Recommended for researchers, statisticians, and data analysts working with group comparisons.

Install

scikit-posthocs on PyPI

pip install scikit-posthocs

uv add scikit-posthocs

poetry add scikit-posthocs

Installing scikit-posthocs

Before you install

Low friction install with six standard scientific dependencies (numpy, scipy, statsmodels, pandas, seaborn, matplotlib). Package is actively maintained with a recent commit on 2026-07-24 and has been in production since 2018-02-01.

License in practice

MIT license (permissive) allows free use, modification, and distribution with minimal restrictions—suitable for academic, commercial, and open-source projects.

Quickstart

pip install scikit-posthocs

import scikit_posthocs as sp
import pandas as pd

# After ANOVA, perform post hoc test on groups
result = sp.posthoc_tukey(data, val_col='values', group_col='groups')

Requires Python 3.9 or later; all six runtime dependencies (numpy, scipy, statsmodels, pandas, seaborn, matplotlib) must be installed.

Verify before relying

• Whether the package supports effect size calculations or confidence intervals alongside p-values.
• Performance characteristics when handling large datasets or many groups.
• Specific visualization capabilities beyond 'significance plots' mentioned in the description.

Package facts

Evidence: scikit_posthocs-0.14.0-py3-none-any.whl

Keywords: statistics, stats, posthoc, anova, data science

Tags

More Information Analysis packages