karateclub

Why we included this project

Karate Club is a Python library for unsupervised learning on graphs, built as an extension to NetworkX. It bundles a wide range of established methods: node- and graph-level embeddings, plus overlapping and non-overlapping community detection, all exposed through a scikit-learn-style fit/predict API that makes swapping techniques a matter of a few lines of code. Because it sits on top of NetworkX, it drops naturally into Python workflows that already use that ecosystem. That combination makes it a good fit for exploratory analysis, feature engineering for downstream classifiers, or research that needs to compare several graph mining approaches quickly without hand-coding each one. The package also ships graph classification benchmark datasets, which gives you comparable starting points for evaluation.