No. 112 of 500 No. 17 in AI for Beginners

The Art of Statistics

How to Learn from Data

By David Spiegelhalter

Print length
448 pages
Publisher
Basic Books
Reading level
Beginner
ISBN-13
978-1541618510
Published
September 3, 2019
Written for
Curious readers, journalists and professionals who want sound judgment about data-driven and algorithmic claims.

Synopsis

David Spiegelhalter, a Cambridge statistician and former president of the Royal Statistical Society, wrote this book to teach statistical thinking through questions rather than formulas. Each chapter opens with a real problem, such as which passengers were most likely to survive the Titanic, whether the serial killer Harold Shipman could have been caught earlier, or whether screening for ovarian cancer saves lives, and works through it using a problem, plan, data, analysis and conclusion cycle known as PPDAC. Along the way he covers summarizing and communicating numbers, populations and measurement, causation, regression, probability, confidence intervals, hypothesis tests and Bayesian reasoning, and closes with chapters on how statistical work goes wrong and how it can improve.

Chapter six speaks most directly to readers interested in AI. Using the Titanic data, Spiegelhalter builds classification trees and regression models, explains overfitting, training and test sets and ROC curves, and lists warning signs of a flawed algorithm, including lack of robustness, ignored statistical variability, implicit bias and lack of transparency. The treatment is brief but grounded, and it shows how machine learning grows out of older statistical ideas. Readers gain the ability to judge headline numbers, separate correlation from causation and recognize when a model's claimed accuracy deserves skepticism. The prose is clear, but the later chapters on probability, p-values and Bayesian inference ask more of the reader than most popular statistics books.

Scorecard

Editorial score

How it scored

Editorial score: 77.8 / 1005 weighted criteria

  • Influence30%

    22.5/30
  • Authority25%

    22.5/25
  • Practical value20%

    14.0/20
  • Currency15%

    11.3/15
  • Readability10%

    7.5/10
  • Editorial score

    77.8/100

How the rank was computed

Rank No. 112 of 500 comes from a composite of 62.3: observable signals 27.3 of 55 points plus editorial review 35.0 of 45 points.

All 11 inputs
InputPoints
Observable signals27.3/55
Reach (Amazon ratings volume)13.3/16.5
Consensus across best-of lists2.8/13.8
Awards and bestseller evidence3.3/8.3
Reader rating, sample-adjusted6.9/8.3
Amazon sales rank1.0/5.5
Recency0.0/2.8
Editorial review35.0/45
Author authority10.1/11.3
Influence10.1/13.5
Practical value6.3/9.0
Readability3.4/4.5
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 statistician's rigorous yet readable guide to learning from data, with a clear-eyed chapter on prediction algorithms and their limits.

Strengths

  • Real-world cases anchor every chapter
  • Authoritative treatment of uncertainty and causation
  • Balanced coverage of frequentist and Bayesian methods

Keep in mind

Later chapters on probability, hypothesis testing and Bayesian inference grow demanding, and machine learning receives a single chapter rather than a full treatment.

Key takeaways

  1. Statistical problems are best tackled through a cycle of problem, plan, data, analysis and conclusion, and many errors originate before any calculation begins.
  2. How numbers are framed, as relative or absolute risk, percentages or counts, changes how audiences perceive the same evidence.
  3. Prediction algorithms are statistical models that must be checked on independent test data and scrutinized for robustness, bias, uncertainty and transparency.
  4. Many published findings fail to replicate because of questionable research practices, which makes skepticism toward single striking results a sound default.

Notable facts

About the author

David Spiegelhalter

Sir David Spiegelhalter is emeritus professor of statistics at the University of Cambridge, where he was Winton Professor of the Public Understanding of Risk from 2007 to 2018. A Fellow of the Royal Society, he led the MRC team that developed the WinBUGS Bayesian software, was knighted in 2014 and served as president of the Royal Statistical Society in 2017 and 2018.

Questions

Is The Art of Statistics good preparation for learning about AI?
Yes. Spiegelhalter devotes a full chapter to algorithms, analytics and prediction, building classification models from Titanic passenger data and explaining overfitting, test sets, ROC curves and algorithmic bias. The rest of the book supplies the reasoning about uncertainty and causation on which machine learning practitioners rely. It is not a technical AI text, but it equips readers to evaluate predictive systems critically.
How does The Art of Statistics compare with Naked Statistics?
Both explain statistics without heavy math, but they differ in depth and tone. Charles Wheelan's Naked Statistics leans on humor and is gentler for complete beginners. Spiegelhalter's book is more rigorous, covers Bayesian methods, reproducibility and prediction algorithms, and draws on his long career in medical statistics. Readers who finish Wheelan and want more substance will find Spiegelhalter the natural next step.
Who should read The Art of Statistics?
It suits general readers, students, journalists, doctors, managers and policymakers who encounter statistical claims and want to judge them. No advanced math is required, though the later chapters on probability and hypothesis testing demand concentration. Practicing statisticians may find the material familiar but useful for teaching, since Spiegelhalter is known for explaining risk and uncertainty to public audiences.

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.