文档介绍：JSS Journal of Statistical Software
February 2009, Volume 29, Book Review 11. /
Reviewer: John Maindonald
Australian National University
Modern Multivariate Statistical Techniques: Regression, Classification and
Manifold Learning
Alan Julian Izenman
Springer-Verlag, New York, 2008.
ISBN 978-0-387-78188-4. 732+xxvi pp. USD .
/~alan/MMST/
This is a near encyclopedic account of topics e generally under the headings of
regression, classification, cluster analysis and low-dimensional representation, albeit with a
strong statistical learning perspective. Graphical displays are used to excellent effect. It is
parable in style and content to Hastie, Tibshirani, and Friedman (2001), due out
shortly in a second edition.
Theoretical demands are more severe than for Hastie et al. (2001). As with that text, it
stays strictly within an independent observations theoretical framework. I could not find any
discussion of issues that arise from mon attempt to generalize beyond the population
that have generated the data.
A high level of theoretical preparation is expected:
As prerequisites, readers are expected to have had previous knowledge of prob-
ability, statistical theory and methods, multivariate calculus, and linear/matrix
algebra. [. . . ] Along with a background in classical statistical theory, it would
also be helpful if the