ABSTRACT
Correspondence analysis (CORA) is an exploratory technique for analysing multi-way frequency tables, that is, cross-classifications of two or more categorical variables. We will focus on the analysis of two-way tables, but the analysis of multi-way tables depends on much the same set of ideas and will be discussed briefly later in the chapter. Like MDS, correspondence analysis aims to convert a table of numbers into a plot of points in a small number of dimensions — usually two. The term correspondence analysis derives from the French Analyse Factorielle de Correspondances which is the term used by Benzecri and others who developed the technique. However, the basic idea is found much earlier in attempts to scale the categories of contingency tables.
