Geometric data analysis

The label Geometric Data Analysis was proposed by P. Suppes of Stanford University for the multivariate statistics framework that Benzécri introduced under the name Correspondence Analysis. The method has steadily gained users and esteem since then. Brigitte Le Roux sets out the complete formal foundation of the approach in linear algebra, which is its most distinctive and far-reaching feature. She also explains how conventional statistical techniques, including Analysis of Variance and Bayesian methods, can be incorporated into it.

Chapter 9, Research Case Studies, is long enough to stand as a book in its own right. It applies the methodology to three substantial datasets drawn from medicine, political science, and education. The educational data come from the computer-based Educational Program for Gifted Youth at Stanford.

The book is aimed at two groups: mathematicians who want to see how mathematics is applied to real problems, and researchers who want to acquire a powerful technique for analysing statistical data.