No. 225 of 500 No. 37 in Machine Learning

Data-Driven Science and Engineering

Machine Learning, Dynamical Systems, and Control

By Steven L. Brunton and J. Nathan Kutz

Published
July 28, 2022
Print length
614 pages
Reading level
Advanced
ISBN-13
978-1009098489
Publisher
Cambridge University Press
Written for
Engineering and physical science students, researchers and practitioners applying machine learning to dynamical systems.

Synopsis

Steven Brunton and J. Nathan Kutz, both at the University of Washington, wrote this textbook for engineers and physical scientists who need to model, predict and control complex systems from data. It opens with the mathematics of reducing high-dimensional data, including the singular value decomposition, Fourier and wavelet transforms, and sparsity with compressed sensing. A second part covers regression and model selection, clustering and classification, and neural networks. The book then turns to its distinctive ground: learning the behavior and governing equations of dynamical systems from measurements, linear and data-driven control, and reduced-order models for simulations too expensive to run in full.

The second edition, published in 2022, adds chapters on reinforcement learning and physics-informed machine learning, along with chapter exercises. Methods are presented with worked examples, many drawn from fluid dynamics and other physical systems, and the companion website supplies lecture videos for each section, homework, data and code in MATLAB, Python, Julia and R. Coverage runs from undergraduate fundamentals to current research methods, so it serves both coursework and self-study. Readers gain a unified toolkit that connects classical applied mathematics to modern machine learning, and a sense of when data-driven models can complement or replace physics-based ones.

Scorecard

Editorial score

How it scored

Editorial score: 80.3 / 1005 weighted criteria

  • Influence30%

    24.0/30
  • Authority25%

    22.5/25
  • Practical value20%

    16.0/20
  • Currency15%

    11.3/15
  • Readability10%

    6.5/10
  • Editorial score

    80.3/100

How the rank was computed

Rank No. 225 of 500 comes from a composite of 55.6: observable signals 19.5 of 55 points plus editorial review 36.1 of 45 points.

All 11 inputs
InputPoints
Observable signals19.5/55
Reach (Amazon ratings volume)8.1/16.5
Consensus across best-of lists2.8/13.8
Awards and bestseller evidence0.0/8.3
Reader rating, sample-adjusted7.4/8.3
Amazon sales rank1.2/5.5
Recency0.0/2.8
Editorial review36.1/45
Author authority10.1/11.3
Influence10.8/13.5
Practical value7.2/9.0
Readability2.9/4.5
Currency5.1/6.8

Each criterion is scored from 1 to 10 and weighted, and the points add up to the editorial score. The rank combines observable signals (55 percent) with this score (45 percent). Read the methodology.

Verdict

A leading textbook joining machine learning to dynamical systems and control, backed by extensive code and lecture videos.

Strengths

  • Unifies applied mathematics, control and machine learning
  • Extensive code, data and lecture videos
  • Updated with reinforcement learning and physics-informed ML

Keep in mind

Requires linear algebra, differential equations and some programming; readers seeking a general machine learning survey or a business perspective will find it too specialized.

Key takeaways

  1. The singular value decomposition underpins much of data-driven modeling, from principal component analysis to proper orthogonal decomposition and reduced-order models.
  2. Sparsity is a modeling principle: compressed sensing and sparse regression let engineers recover signals and identify governing equations from limited measurements.
  3. Dynamic mode decomposition and Koopman operator methods let practitioners extract linear, predictive models of nonlinear systems directly from data.
  4. Physics-informed machine learning embeds known physical laws and symmetries into models, improving accuracy and generalization when data are scarce.

Notable facts

  • Kianoosh Yousefi of the University of Texas at Dallas called it one of the best textbooks in the field and said he was adopting it for his course; Shirley Ho of the Flatiron Institute and NYU also endorsed it. Source (opens in a new tab)
  • The second edition, published by Cambridge University Press on 5 May 2022, added chapters on reinforcement learning and physics-informed machine learning. Source (opens in a new tab)
  • Kutz and Brunton were named director and associate director of the UW-led NSF AI Institute in Dynamic Systems, one of 11 new NSF AI research institutes announced in 2021. Source (opens in a new tab)

About the author

Steven L. Brunton and J. Nathan Kutz

Steven L. Brunton is the Boeing Professor of AI and Data-Driven Engineering in mechanical engineering at the University of Washington, a PECASE recipient and an American Physical Society Fellow. J. Nathan Kutz is the Robert Bolles and Yasuko Endo Professor of Applied Mathematics at UW. Together they lead the NSF AI Institute in Dynamic Systems, and each has written three textbooks.

Questions

What is Data-Driven Science and Engineering about?
It is a graduate-level textbook on using data and machine learning to model, predict and control physical and engineered systems. It covers dimensionality reduction with the singular value decomposition, sparsity and compressed sensing, regression, classification and neural networks, then data-driven dynamical systems, control theory, reduced-order models, reinforcement learning and physics-informed machine learning, with code and lecture videos online.
Who should read Data-Driven Science and Engineering?
Advanced undergraduates, graduate students and working engineers in mechanical, aerospace, electrical and applied mathematics fields who want to apply machine learning to systems that change over time. Readers need linear algebra, calculus and differential equations, plus some MATLAB or Python. Those looking for a general introduction to AI or for business strategy will find it far more technical than they need.
Should I read the first or second edition of Data-Driven Science and Engineering?
Choose the second edition, published by Cambridge University Press in 2022. It keeps the core material on the singular value decomposition, sparse methods, dynamical systems and control, and adds chapters on reinforcement learning and physics-informed machine learning plus chapter exercises. Supporting code, data and lecture videos are available on the companion website, which makes the newer edition well suited to self-study.

Sources

As an Amazon Associate, Best AI Books earns from qualifying purchases. Best AI Books is sponsored by Iternal.ai, whose founder and CEO, John Byron Hanby IV, wrote The AI Strategy Blueprint, our #1 ranked book. How we rank books.