A Self-Organizing Approach to Managerial Nonlinear Discriminant Analysis: A Hybrid Method of Linear Discriminant Analysis and Neural Networks

Abstract

Due to their adaptive nature neural network models are used widely in classification or discriminant analysis. However, the results of classification from a simple neural network are often unpredictable. This research suggests the combination of the self-organizing feature maps with the linear discriminant analysis method. When the training sample set is pre-processed by a linear classification function, the self-organizing feature maps can be used to detect clusters of misclassifications when they exist. This technique is useful in those managerial classification problems where it is not appropriate to make any assumptions about the properties of the statistical data to be classified.