Tanagra Ricco Rakotomalala 1 1 Introduction

However, there are strong connections between these approaches when we deal with a binary target attribute. In this particular case, we can even recreate the outputs of the linear discriminant analysis with a linear regression program (Bishop, 2007, pages 189 190; Duda et al., 2001, pages 242 – 243; Huberty et Olejnik, 2006, pages 353 – 355; Nakache et Confais, 2003, pages 14 – 16; Saporta, 2006, pages 451 – 452; Tomassone et al., 1988, pages 36 – 38). Unfortunately, if the various references show the connections between the matrix expressions, some explaining the transition formulas, no one details the calculations on a numerical example, making the demonstration too abstract. We perceive badly the real scope of this equivalence. By searching on the Web (in English and French), I ended up finding a detailed example that highlights the relationship. The coefficients of the linear functions from the two approaches are proportional, alas, without that the author details the mathematical expression of the ratio between the coefficients (Desbois, 2003; page 31).