Linear Discriminant Analysis Algorithm

Linear discriminant analysis, or LDA, is a generalization of Fisher's linear discriminant used to find a linear combination of features that characterizes or separates two or more classes of objects, and the resulting combination can serve as a linear classifier or, more commonly, for dimensionality reduction before later classification. It is closely related to principal component analysis in that both look for linear combinations of variables that best explain the data, but LDA explicitly models the difference between classes of data while PCA does not take class membership into account at all. Discriminant analysis requires the class groups to be known in advance, unlike cluster analysis, and each case must carry a score on one or more continuous predictor measures alongside a group label.