OpenMOA

OpenMOA is a Python library designed for practical, utilitarian online learning tasks. It extends the CapyMOA framework to support real-world challenges in data stream processing, including evolving feature spaces, concept drift, and heterogeneous streaming data.

Key contributions:

• Architected and developed core modules for online learning under non-stationary environments.
• Extended CapyMOA with support for data streams with evolving and heterogeneous feature spaces.
• Implemented efficient algorithms for handling concept drift detection and adaptation.

Manuscript: OpenMOA: A Python Library for Utilitarian Online Learning — Zhili Wang, Heitor M. Gomes, Yi He. Under review at JMLR MLOSS.

Website: openmoa.net