Computer Age Statistical Inference
The twenty-first century has seen a breathtaking expansion of statistical methodology, both in scope and in influence. 'Big data', 'data science', and 'machine learning' have become familiar terms in the news, as statistical methods are brought to bear upon the enormous data sets of modern science and commerce. How did we get here? And where are we going? This book takes us on an exhilarating journey through the revolution in data analysis following the introduction of electronic computation in the 1950s. Beginning with classical inferential theories - Bayesian, frequentist, Fisherian - individual chapters take up a series of influential topics: survival analysis, logistic regression, empirical Bayes, the jackknife and bootstrap, random forests, neural networks, Markov chain Monte Carlo, inference after model selection, and dozens more. The distinctly modern approach integrates methodology and algorithms with statistical inference. The book ends with speculation on the future direction of statistics and data science.
Stats for nerds
14recommend
20mentions
17people
2016–2026years
80% recommend20% neutral
Mentions per year
What Hacker News says
- Get a book on applied statistics, such as
Think Stats: Probability and Statistics for Programmers (https://allendowney.github.io/ThinkStats/)
Computer Age Statistical Inference (https://hastie.su.domains/CASI/)
Statistical Rethinking (https://xcelab.net/rm/)
An Introduction to Statistical Learning (https://www.statlearning.com/)
Obviously maths is going to be involved to do the subject justice. These recommendations are more about applied statistics, but that's the foundation. From there it is a small transition to statistical process control.
- Simple formulas only work because the models themselves for those polls are incredibly simple and adding a bit more complexity requires a lot of tools to compute these uncertainties (this is part of the reason you see probabilistic programming so popular for people doing non-trivial polling work).
There are no simple approximations for a range of even slightly complex models. Even some nice computational tricks like the Laplace approximation don't work on models with high numbers of parameters (since you need to compute the diagonal of the Hessian).
A good overview of the situation is covered in Efron & Hastie's "Computer Age Statistical Inference".
- Good review article. I always enjoy browsing the references, and found "Computer Age Statistical Inference" among them. Looks like a good read, with a pdf available online.
- Elements of Statistical learning is from the same authors (mostly) as ISLR but it's longer and goes a bit deeper on the theory. Also, Computer Age Statistical Inference which is newer and covers more material.
Other big titles - Pattern Recognition and Machine Learning (Bishop), Machine Learning: A Probabilistic Perspective (Murphy) - these last two weren't for me but different people respond to different approaches so check them out especially if the first two books don't click for you.
- I've been brought up in a strictly classical/frequentist/objectivist school of inference and testing. The last few months, I've begun expanding my perspectives into subjective probability, decision analysis, Bayesian methods, causality analysis, and some information theory.
These things have helped me understand probability and statistics a lot better. Some examples of books that have been useful include,
- Understanding Uncertainty (Lindley 2009),
- Theory of Probability (de Finetti 1975),
- Foundations of Statistics (Savage 1953),
- Computer Age Statistical Inference (Efron),
- Causality (Pearl 2011),
- Trustworthy Online Controlled Experiments (Kohavi et al.)
However, most of this is about one "person" dealing with "the world", which in turn is neutral. I want to learn more about the situation where two "people" are in an antagonistic/competitive relationship.
I mention these to give an idea of what sort of book I'd prefer: I want to go into technical detail, but I don't need everything to be proved from axioms. I will eventually use the tools I learn in practise, but I do want an intuitive sense of what they mean and how they are constructed, and I'm prepared to work a little bit for it.
If I can't get a better suggestion here, von Neumann's classic on the subject is where I'll go.
- Stein's paradox [1]. Quite well explained in the book "Computer Age Statistical Inference", available online [2]. Cute quote from that book: "The James–Stein theorem came as a rude shock to the statistical world of 1961".
Related
Statistical Rethinking
81 recommend · 113
Introduction to Statistical Learning
85 recommend · 120
Bayesian reasoning and machine learning
15 recommend · 22
The Elements of Statistical Learning
144 recommend · 266
Doing Bayesian Data Analysis
22 recommend · 33
Machine learning
15 recommend · 31
