Introduction

Machine Learning for Data Streams: With Practical Examples in MOA (2017) provides a comprehensive introduction to learning from continuous, fast‑arriving data streams—an increasingly important challenge as modern systems generate massive, real‑time information flows. Unlike traditional batch learning, data‑stream environments require algorithms that can process each instance once, adapt to evolving patterns, and operate efficiently under strict memory and time constraints. The book offers both the theoretical foundations and practical tools needed to build such adaptive systems, with a strong focus on real‑world applications.

Outcomes

The book delivers a detailed framework for understanding and implementing machine‑learning methods tailored to data streams. It explains core concepts such as incremental learning, concept drift, evaluation under streaming conditions, and the limitations of storing or revisiting past data. A major contribution is its practical orientation: the authors provide hands‑on examples using the MOA (Massive Online Analysis) software, demonstrating how to build, test, and compare streaming algorithms in realistic scenarios. The text also introduces key families of stream‑learning methods, including decision trees, ensemble techniques, drift detectors, and adaptive classifiers showing how they can be combined to achieve robust performance in dynamic environments. Overall, the book equips readers with both the conceptual understanding and practical skills needed to design effective real‑time learning systems.

This work directly involves Albert, Geoff, and Bernhard, who along with their co‑author Ricard Gavaldà, have contributed significantly to the development of adaptive, scalable machine‑learning methods. Their expertise continues to support TAIAO’s mission to advance AI for complex, continuously evolving environmental and ecological data streams.

Bifet, Albert, Ricard Gavaldà, Geoff Holmes, and Bernhard Pfahringer. Machine Learning for Data Streams: With Practical Examples in MOA. MIT Press, 2017.