Customer Reviews

Discrete Choice Methods with Simulation

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

 

 

If I could give this book six stars I would. It's simply one of the best statistics books I've ever read.

This book is very well-written by one of the experts in the field. It covers logit models and the various generalizations (GEV, mixed logit, probit, etc.) in detail, along with a thorough discussion of modern estimation of these models. What I find most useful about it is that the words-to-equations density is highly favorable. The equations you need are there, but the words you need are there too, making sure you understand the model assumptions inside and out. Each equation is explained thoroughly and the surrounding discussion probes the model to bring the reader to a critical understanding of what exactly is implied by the model. Too often complex statistical models are treated in a "black box" fashion. The dirty little secret is that it's easier for the author to do this. Train doesn't take the easy way out. The fact that his web site has truly excellent support--including a large number of webinars in addition to the more usual papers, software, etc.--makes this book a doubly valuable item. See http://elsa.berkeley.edu/~train/distant.html for even more.

-Enjoyable read
-Does not assume PhD level of econometrics
-Explanations are clear and concise

Actually, an advanced undergrad may find this book usefull as well.
Is this the best discrete choice methods book ever published?
Let y = 1 or 0 where 1 = yes, this is the best and 0 = No, the book is not the best. Also let P = Prob (y=1). My results show P = 0.98 (see forthcoming Econometrica article).
While finishing a Master's Thesis in applied econ which focuses on a multinomial logit model, I have sought good info on this topic. Not having gone through the rigors of a PhD program, I have gone through many of the important books and articles which address discrete choice modeling methods some of which address a post-doc audience only. Dr. Train's is by far the best I have encountered. His explanations are concise yet not too dense (see Amemiya). I first encountered many of the concepts in other publications but did understand them until reading Train's book. In my opinion Train has that rare quality of being, not only an exceptional economist, but quite enjoyable to read.

This book is one of the best for introduction to Discrete choice models. I had been using Ben-Akiva & Lerman, but feel this book should be read along with that one for a better understanding of choice models. Also, Train covers recent advances in the field and provides a good introduction to Halton draws. He really makes sure you get the concepts and the online lecture series are really excellent.

As a previous reviewer has already stated, this may quite possibly be the best statistics/econometrics book. It's not a textbook with end-of-chapter problems covering hundreds of topics. It's more of an in-depth narrative on the single important subject of discrete choice models (e.g., logit, probit, and many advanced extensions). The reader should have a background in statistics at the level of Casella & Berger's text. So this book is great for researchers and graduate students in math, stats, engineering, and economics, but is probably too advanced for those in psych, polysci, and sociology. If you're even thinking about doing some sort of discrete choice analysis, this book is a must read. It's extremely insightful, and Train's ability to explain everything in both mathematical and non-technical terms is astonishing. Everytime I think to myself "I wonder what situation would fit here," Train never fails to give a concise, informative example. If his teaching style is anything like what is in this book, it may be worth taking a leave from UCSD to visit Berkeley for a semester just to take his class.

Enough math to solidify the explanation and enough text to make it readable. Very well done.