People sometimes ask me if I know of a good book on a particular technical topic. Here are some that I like.

Statistics and Machine Learning

  • The Elements of Statistical Learning

    Hastie, Tibshirani, and Friedman
    • A comprehensive text on bread-and-butter statistical methods used in data science
    • Emphasizes the mathematical rationale behind methods for predictive modeling, dimensionality reduction, and function approximation
    • Great as an introduction and as a reference: I got my copy in 2016 and still pick it up off my bookshelf

  • Reinforcement Learning

    Sutton and Barto
    • An introduction to models that learn to make predictions by interacting with their environments
    • Written for engineers, but still interesting for neuroscientists and psychologists
      • I distilled some of the key concepts for systems neuroscientists in a seminar series in 2019; you can view the slides here

  • Deep Learning

    Goodfellow, Bengio, and Courville
    • A survey of popular black-box methods for function approximation and the heuristics behind them
    • Buy this book if you want a lightly mathematical survey of common neural network architectures and training algorithms
    • Buy a different book if you want to get a machine learning pipeline off the ground as fast as possible

Writing Code

  • Clean Code

    Robert C. Martin
    • Could have been titled “How to Write Code that Other People Can Understand”
      • This is a severely undervalued skill in science, especially condering that “other people” is a category that will come to include you in three weeks time
    • Written in a slightly argumentative style and full of admonishments that will raise eyebrows (my favourite being that the ideal function is one line long and takes no arguments), but concrete, persuasive, and well worth a read

  • Effective C++

    Scott Meyers
    • Useful to read for its concrete treatment of object oriented programming even if you don’t write C++
    • As a person who lives mainly in Python, this gave me a more thorough appreciation for all the low-level high-performance headaches I don’t have to deal with on a daily basis

  • The R Inferno

    Patrick Burns
    • The first and probably funniest book I’ve read about programming
    • From the abstract: “If you are using R and you think you’re in hell, this is a map for you.”
      • Written for scientists and statisticians struggling with programming, not programmers struggling with statistics
    • You might need to read this book if you don’t know why 1.1 + 2.2 != 3.3
      • It’s free from the author

Neuroscience

  • Neuronal Dynamics

    Gerstner, Kistler, Naud, and Paninski
    • An introduction to mathematical models of biological neurons written mainly for physicists and engineers
    • Emphasis on GLM-based models of single neurons