Discriminant Analysis Programme

Discrimination and Classification of data to and from groups with classical/robust estimation

Description

The purpose of Discriminant Analysis Programme (DAP) is to facilitate discrimination and classification of (to be) grouped data with robust estimation- and modeled structures for the covariances in a one-go software. The robust estimation methods are the S-estimator and Donoho-Stahel estimator. The included covariance structure models are Common Principal Components, Proportional, classical Quadratic and Linear ones; Hypothesis Testing is performed for these fitted models except for arbitrary covariances. The Discriminant Rules are found and the Classification Rules Coefficients are computed after the given training data sample and used to classify the classification sample data, if provided. They are also used to find malclassified data elements by Cross-Validation (Leave-One-Out) method of the training sample, being recomputed for each element.
It includes complementary graphical outputs for bivariate data such as normality plots and group separation.

• MATLAB release: MATLAB 6.5 (R13)