Customer Reviews

An Introduction to Statistical Modeling of Extreme Values

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

 

 

I recently used the software accompanied to this book kindly made available by the author and was led to know more about this book and the author's other works. I like what I saw and think the author has done a supeb job in explaining the difficult theory in plain language and in the context of data analysis. Thus it is an "action" book instead of the "just theory" as with most other books. The book provides a balanced treatment of different approaches to extreme value analysis. Personally I prefer the generalized Pareto approach, though theoretically the point process approach may be very neat, if it can be realized.
I think extreme value theory in general is an important statistical area, since in practice one may be forced to deal with analyzing extreme events, such as in financial engineering, environmental or climate analysis, or network design. I wholeheartedly recommend this book for anyone who want to learn this area from one of the leading researchers.

This book is the most current text available on the theory of extreme values. The author eloquently provides us with an understanding of the theory and it vast applications. It is intended for researchers students and practitioners. So it provides an in-depth account of the theory with many real world examples. It contains an excellent up-to-date bibliography. Important theorems are presented with their implications but without mathematical proofs. Computations are done in SPlus. The author provides an appendix on computational aspects that tells the reader where to go to download examples and find the SPlus functions that are used.
Topics include classical extreme value theory and models, threshold models, extremes in dependent stationary cases, extremes for some nonstationary stochastic processes, the point process approach, multivariate extremes and some special topics including extremes in spatial processes and the Bayesian approach to extremes (with examples employing MCMC methods). There has been a surge in publication of texts on extremes, some theoretical and some applied. This text and a few others are good at intrroducing the topic to the uninitiated and placing emphasis on applications for the practitioners.

This book is a good blend of necessary theory supplied with numerical examples (including financial time series). Despite being short, it is very informative and reads easily. However, if you want more than just a good introduction you have to move forward in time - there's been much progress in Extreme Value field since 2001.

For instance, you may check out "Extreme Values..." and "Heavy-tail phenomena..." by Sidney Resnick.

Stuart Coles, who is well published on the open literature, has delivered this practical text on extreme values statistics by providing extreme values theory in a simplified manner with worked examples. Further, Coles has removed much of the complicating aspects (long mathematical proofs and overly complex notation) typical of statistical literature. The text provides any engineer or scientists with the tools required to complete routine extreme values analysis.