Discriminatory Analysis

The method of finding a linear combination of variables that best separate one or two classes. Related to the group of boundary methods. The method is based on the assumption that boundaries between classes are approximated by linear functions. Then the task is to evaluate parameters of these functions.

The method is widely used to solve classification problems, pattern recognition, and dimensionality reduction of input data. Although the method deals with information that determines whether an object belongs to one of the classes, it is not a classifier itself but used as part of the linear classification model.