Machine Learning on the Fly: A Hands-On Tutorial for Streaming Data

Data stream learning is an emerging machine learning paradigm designed for environments where data arrive continuously and must be processed in real time. Unlike traditional batch learning, which assumes access to a fixed dataset, stream learning addresses the unique challenges of non-stationary distributions, bounded memory, and strict computational constraints. These challenges are increasingly relevant across domains such as IoT, finance, cybersecurity, and environmental monitoring, where timely and adaptive decision-making is essential. This tutorial introduces key concepts and techniques in data stream learning, blending foundational theory with practical demonstrations. It features CapyMOA, an open-source library that provides efficient algorithm implementations through a high-level Python API. We demonstrate the use of this tool through practical examples, with all source code available at https://github.com/adaptive-machine-learning/CapyMOA, and supporting tutorials and installation guides accessible at https://capymoa.org/.