Numerical Methods of Statistics (Cambridge Series in Statistical and Probabilistic Mathematics, Series Number 7)

This book explains how computer software is designed to perform the tasks required for sophisticated statistical analysis. For statisticians, it examines the nitty-gritty computational problems behind statistical methods. For mathematicians and computer scientists, it looks at the application of mathematical tools to statistical problems. The first half of the book offers a basic background in numerical analysis that emphasizes issues important to statisticians. The next several chapters cover a broad array of statistical tools, such as maximum likelihood and nonlinear regression. The author also treats the application of numerical tools; numerical integration and random number generation are explained in a unified manner reflecting complementary views of Monte Carlo methods. The book concludes with an examination of sorting, FFT and the application of other "fast" algorithms to statistics. Each chapter contains exercises that range in difficulty as well as examples of the methods at work. Most of the examples are accompanied by demonstration code available from the author's home page.

This book explains how computer software is designed to perform the tasks required for sophisticated statistical analysis. The first half of the book provides a basic background in numerical analysis emphasizing issues important to statisticians. The next several chapters cover a broad array of applications, such as maximum likelihood and nonlinear regression, numerical integration and random number generation, Monte Carlo methods, sorting, FFT and the application of other "fast" algorithms to statistics. Numerous examples are accompanied by demonstration code available on a floppy disk included with the book. This graduate text will also be a valuable reference for statisticians, mathematicians, and numerical analysts.