CAIMAN brothers: A family of powerful classification and class modeling techniques

CAIMAN (Classification and Influence Matrix Analysis), a new classification technique, is here analyzed and modified to produce a number of possible classification and class modeling techniques with good performances in that regards both the prediction ability and the efficiency of the class models. These techniques are based on the addition to the original data matrix of the matrix of the Mahalanobis distances from the class centroids (or of the leverages, or of other distances). Then, the classical techniques of classification and class modeling are applied to the blocks of the predictors (original, added), separately or after fusion. © 2009 Elsevier B.V. All rights reserved.