Implementation of linear and quadratic discriminant analysis incorporating costs of misclassification

Discriminant analysis plays an important role in biological and medical research. In practice, standard linear and quadratic methods are often applied which assume equal costs of misclassification. However, there can be situations where misclassifications between certain groups may be more serious than between other groups. Such considerations can be taken into account by using classification methods which incorporate misclassification costs. The widely applied statistical packages BMDP, SAS, and SPSS do not offer the possibility of using unequal misclassification costs for discriminant analysis with more than two groups. In this paper a menu-driven, user-friendly PC program written in Borland Pascal is introduced which performs linear and quadratic discriminant analysis for g > or = 2 groups allowing for the incorporation of misclassification costs.