Customer Reviews

Bayes and Empirical Bayes Methods for Data Analysis, Second Edition

5 star:		(2)
4 star:		(0)
3 star:		(1)
2 star:		(0)
1 star:		(0)

 

 

This book features a deep and focused lesson on Bayes and Empirical Bayes Methods. It goes through the key topics as conjugate priors, MCMC methods (non iteratives and iteratives as the well known Gibbs samplining and metropolitis hastings algorithms), model selection methods (as bayes factor) and issues related as model robusteness.
The Approach is increasingly formal and deeply complex, allowing for getting the basics or diving into more complex knowledge according to your former background. You need at least a good understanding of Frequentist statistic to be able to follow the reasonings. Each chapter allow you to stop at some point without losing the thread. Last part of the book is in fact deep knowledge demanding.
The most interesting point of this book according to my very limited statistics background is that it makes good comparations with the frequentist approach (classical approaches as confidence intervals and point estimators), checking performance of either method. Even, it features some combination of both approaches getting some bayessian intervals.
As a negative point, I would say that examples are hard to follow for someone with limited bakground and too much complex. They really do not clear me up enough.
All in all, is a very profitable book for jumping into bayesian methods.

Bayesian Statistics is being use more and more these days because the amazing advances in computational speed allow the use of computer-intensive methods to calculate Bayesian posterior distributions using more realistic prior distribution.

The first edition of this book was a well-written primer on Bayesian methods and the more "objective" empirical Bayes methods. The second edition adds much more on Gibbs sampling and algorithms such as Metropolis-Hastings that enable statisticians to produce realistic Bayesian results using the Markov Chain Monte Carlo techniques. Although some instructors do not sind it to be the best text for a course on Bayesian methods, it is a valuable reference yexy for statisticians and is well-suited for a graduate level text.

For a first course that includes in greater detail examples and applications I would prefer Jeff Gill's book which although written for the audience of social scientists, it can be used in other disciplines as well.

We used this book for our intro to Bayesian statistics class at SDSU. I thought it was more like a technical manual for how to do Bayesian statistics, rather than a good introductory textbook. Recommended for researchers who want to know the nitty-gritty of MCMC and the like. Not a good textbook for a first course in Bayesian statistics. To understand what was going on in class I used Lancaster, "An Introduction to Bayesian Econometrics". Much better intuitive explanation of what is going on.