Handbook of computational statistics. Concepts and methods. (Q4818643)

As the editors state, the ``purpose of this handbook is to provide a survey of the basic concepts of computational statistics''. The book contains 35 expository articles on various topics from the field of computational statistics written by experts in those fields. Each of the contributions describes on about 30 pages the basic ideas, fundamental concepts and current research topics in the relevant area, concluding with a long list of references for further reading. The articles are written for a general audience with background knowledge in statistics and computational methods and an interest in combining these fields in an up-to-date presentation of computational statistics. In the first part of the book, the editors give a historical outline of the evolution of the field, including lists of relevant scientific groups, conferences and journals. Following their argumentation, modern computing technology allows much more involved methods to be actually computed with a standard computer. This leads to questions of data representations, efficient data handling and efficient algorithms, up to computer graphics and software engineering. To make these developments available to researchers in statistics, a new area of research evolved. Statistical computing has thus the aim to provide methods which enable the use of modern computational techniques for problems of statistical inference. Two large chapters combine articles on statistical computing and statistical methodology, respectively. The chapter on statistical computing contains contributions on questions of computing with electronic computing devices. Although the focus is on the use of computers for statistical computations, some of the articles are of interest for every user of a computer as tool to calculate. Topics covered in this chapter are, among others, representations of numbers in computers, basic algorithms, random number generation, and simulation of random processes, widely used algorithms like the EM algorithm and transformation algorithms, as well as some concepts of computer graphics and related fields. The chapter on statistical methodology deals with statistical methods which are computationally intensive. From the point of view of methodology of statistical computing, inference based on asymptotic analysis is replaced by inference based on computational analysis. The chapter starts with articles on cross-validation, resampling and simulation. A larger part then deals with model choice. The underlying paradigm is formulated by the editors as ``looking at data through a variety of models''. Estimating multivariate probability density functions and methods for analysis of relations among several variables are treated in the following sections. Further issues are robustness of statistical methods, Bayesian analysis and survival analysis. The chapter closes with introductions to data mining, tree-based methods, support vector machines, and ensemble methods. The last chapter includes descriptions of five application fields to show the impact of computational statistics in selected fields of modern scientific research. There are contributions dealing with risk calculation, econometrics, statistical geometry of biomolecular structure, functional magnetic resonance imaging, and network intrusion detection. The book is accompanied by an ebook version which can be downloaded onto a local computer. It enables to directly use links to internet sources all over the world.