Bayesian Book Reviews Blog for Management and Business Professionals

Bayesian Modeling and Computation in Python

Osvaldo A. Martin, Ravin Kumar, and Junpeng Lao

2021, CRC Press.


Short take - buckle up as the authors have you code your own Metropolis sampler on page 6! We also learn about maximum entropy priors and Bernardo reference priors in Chapter 1. Let's go! I'm in Chapter 2 but my view is this book is for a business/management professional/academic who has very solid grounding in Bayesian methods and Bayesian practice and who wants to migrate to Python from a more traditional Bayesian environment (WinBUGS, STAN, R, JAGS).

Long take - I picked this book up since it is a new book in my field which I came across on Amazon. I do not have any Python experience but I would be interested to learn Python programming with respect to my research interests. The good news for you, the prospective reader, is that the entire book is freely available in PDF and web formats at the authors' site. I was not aware of this at my time of purchase but you can get this book for yourself on any device at any time. At the time of my writing, the instructions for installing of the programming environment were two lines of Conda code. I downloaded and set up Python and Anaconda on two separate machines but could not get these lines of code to do anything. The book also says that the book runs in Google Colab. I am sure it does but again the provided code gave an error in Colab. The Github page reads "More instructions to come soon" in the Environment Installation section time of writing (href=https://github.com/BayesianModelingandComputationInPython/BookCode_Edition1#environment-installation).

Having read through the first several chapters of the book and the authors' biographies, it seems this book is an extended manual that corresponds to multiple computation packages of the authors' creation. If you already use Python and the authors' packages, this book would likely make for some interesting reading. Like any software manual, if you can't run the software then part of the value of reading a book goes away. This is the only Bayesian book I have had to end early due to software. I will revise my short take to say this book is for a business/management professional/academic who has very solid grounding in Bayesian methods and Bayesian practice and who already uses Python and the authors' packages and would like a detailed treatment. Again the good news is that you can see if this book fits your needs by getting it for free at the authors' site. The book is attractively laid out with nice use of color, formatting and graphics so it is a bit of a pity to have to end early. It would have been nice if the authors had an entryway for readers with little to no Python experience. Another possibility is to teach yourself Python and Bayesian modeling elsewhere, and then when you are ready to use the authors' packages, return to this book. Have a look at the book on Amazon here.

Reviewed July 2022.




Back to Gene Hahn's book blog.