Computing Reviews
Today's Issue Hot Topics Search Browse Recommended My Account Log In
Review Help
Search
Machine learning for data streams : with practical examples in MOA
Bifet A., Gavaldà R., Holmes G., Pfahringer B., The MIT Press, Cambridge, MA, 2017. 288 pp. Type: Book (978-0-262037-79-2)
Date Reviewed: Dec 13 2018

Data streams are everywhere. Sensors, people, and several applications continuously produce data streams, that is, related data items in temporal order, like the ones in financial markets or social media. They are also referred to as big data due to their high volume, high variety, and high velocity. Data stream processing algorithms involve non-terminating computations that only use members of the most recent time window and aim to generate decisions or recommendations without a delay. The consumers of these outputs, such as the ones related to the stock market, are humans, algorithms, or both.

This book presents various algorithms used in data stream mining. Its authors are well-known scholars in the community. The presentation is based on massive online analytics (MOA), an open-source project from the authors and some other contributors. Researchers regularly use MOA; it is a highly appreciated tool that advances the field of data stream mining. It provides not only various algorithms, but also synthetic stream generation options that enable tests under various conditions that may be difficult to identify or validate in real data. Users can extend the framework to accommodate new algorithms.

The book has three parts, 15 chapters all together. Part 1 is an introduction to stream mining applications. Here, the authors explain how real-time data stream analytics work, including applications and challenges. Chapter 2 follows with widely known mainstream data mining methods such as majority class classifier, lazy classifier, naive Bayes, and others. This brief but clear introduction is a good starting point for data stream analysis. In chapter 3, MOA software is introduced mainly with its graphical user interface (GUI). There are also some command-line interface (CLI) instructions for simple learner settings without proper command and parameter documentations. This unfortunately leaves very little room for individual understanding and implementation; furthermore, a similar approach to examples and sample applications is followed throughout the book.

Part 2 is the longest section. Chapters 4 to 10 introduce stream mining methodologies in detail. The theoretical aspects of each methodology are well explained and easy to follow. These formulations are especially useful for practitioners and researchers who would like more than just a conceptual understanding.

Part 3 is on the MOA software. Chapter 11 explains the MOA ecosystem, installation, and other useful stream analysis software. The following two chapters are on the GUI and using the command line. Chapter 14 very briefly introduces the MOA application programming interface (API). The book ends with chapter 15, “Developing New Methods in MOA.”

The book’s 262 references are from more than 190 different researchers. It also contains a rich index; however, there is no author index.

One area of improvement for the book is its lab sessions: they are not very “practical,” as suggested by the book’s title. Researchers tend to develop their own models using different APIs, but the book does not offer much help here. It would be nice to have a brief theoretical explanation followed by both API and GUI examples at the end of each chapter, so that readers can immediately practice what they have learned. The CLI commands also need more detail. It would be nice to include such examples earlier in the book, too, for example, at the end of Part 1.

The book may have some deficiencies; however, it is a great source of information on stream mining and analysis for anybody working on this topic, from beginners to advanced learners. All in all, it is a welcome addition to the literature. I would like the next edition to include new developments in this dynamic and evolving research field with important practical implications.

More reviews about this item: Amazon

Reviewer:  F. Can Review #: CR146343 (1902-0017)
Bookmark and Share
 
Learning (I.2.6 )
 
 
Machine Translation (I.2.7 ... )
 
Would you recommend this review?
yes
no
Other reviews under "Learning": Date
Learning in parallel networks: simulating learning in a probabilistic system
Hinton G. (ed) BYTE 10(4): 265-273, 1985. Type: Article
Nov 1 1985
Macro-operators: a weak method for learning
Korf R. Artificial Intelligence 26(1): 35-77, 1985. Type: Article
Feb 1 1986
Inferring (mal) rules from pupils’ protocols
Sleeman D.  Progress in artificial intelligence (, Orsay, France,391985. Type: Proceedings
Dec 1 1985
more...

E-Mail This Printer-Friendly
Send Your Comments
Contact Us
Reproduction in whole or in part without permission is prohibited.   Copyright 1999-2024 ThinkLoud®
Terms of Use
| Privacy Policy