Role: Co-developer Org: Open Science Laboratory (OpenSciLab) Adoption: Added to conda-forge. Adopted as a dependency by Keras and TensorBoard. Used in 200+ research papers, mostly medical, for comparing classification models. Funding: NLnet Foundation Grant — PyCM — funded by NGI0 Commons for distance similarity and benchmarking (Feb 2025), and Trelis AI Micro-Grant — PyCM — a development micro-grant to build a RESTful API for PyCM (Sep 2024).
Summary
PyCM evaluates classifiers after the fact: it works from the predictions a model has already made and supports most per-class and overall statistic parameters you might ask for. It is built for the case where accuracy alone doesn’t tell you anything useful and you need the long list of metrics instead — data scientists who need a broad array of measures to accurately evaluate a wide variety of classifiers, rather than a single headline number.
Links
Related
A Literature Review on Rater Agreement Metrics — technical report supporting the library. O — sibling ML tooling.