Statistical tests that can be used to make inferences about“bad data,” that is, heavy-tailed distributions, arediscussed in this valuable guide. Ironically, the editors indicate thatmuch of the trepidation surrounding the analysis of skewed data isunnecessary. Several tests exist to handle such data. It was simply amatter of researching options. This search, while not easy, wasstraightforward, and the editors and contributors have done a marvelousjob.

The text begins with a section on applications. The applicationscited deal with the Web, structural modeling of network traffic, andeconomic issues of finance and risk management.

The second section deals with time series. There are six papers inthis section. Part 3 of the book deals with heavy-tail estimation. Part4 deals with regression. It consists of two papers, on bootstrappingsigns and permutations for regression with heavy-tail errors, and onlinear regression with stable disturbances.

Part 5 has two papers on signal processing. Part 6 consists ofthree papers on model structures, which cover subexponentialdistributions, the structure of stationary stable processes, and thetail behavior of some shot noise processes.

The text concludes with a section on numerical procedures, whichcould be of particular interest to any reader interested in heavy-taildistributions. It includes papers on numerical approximations of thesymmetric stable distribution and density; a table of maximally skeweddistributions; multivariate stable distributions; and univariate stable distributions. 

Obviously, the book is not necessarily meant for recreationalreading, but professional statisticians and appliedmathematicians will find it fascinating for its presentation anddocumentation of versatile methods for handling troublesome data. Thereis no formal reference section. Each paper concludes with its ownreferences.