ClusterEval 0.2

External clustering evaluation

ClusterEval: External Clustering Validation by the Homogeneity-Parsimony Trade-Off

ClusterEval is a lightweight software package for external clustering validation.

It provides reference implementations for the homogeneity-parsimony scores proposed in Tiffeau-Mayer 2026. Evaluation of clustering quality on these two objectives provides a unified framework for assessing clustering agreement with ground truth class labels.

The scores are implemented to be compatible with evaluation metrics defined in Scikit-learn's sklearn.metrics to allow easy replacement within existing pipelines.

Installation

ClusterEval can be installed via pip:

pip install clustereval

The package depends on numpy and scikit-learn.

Documentation and examples

API documentation is hosted on readthedocs. You can also create a local copy of the API documentation by running:

make html

in the docs folder.