ABSTRACT
Discriminant analysis, often also referred to under the more general notion of pattern recognition, answers the question of which category an observed item is most likely to belong to. A typical application in music is attribution of an anonymous composition to a time period or even to a composer. Other examples are discussed below. A prerequisite for the application of discriminant analysis is that a “training data set” is available where the correct answers are known. We give a brief introduction to basic principles of discriminant analysis. For a detailed account see e.g. Mardia et al. (1979), Klecka (1980), Breiman (1984), Seber (1984), Fukunaga (1990), McLachlan (1992) and Huberty (1994), Ripley (1995), Duda et al. (2000), Hastie et al. (2001).
