Classification of MMPI profiles of patients with mental disorders - Experiments with attribute reduction and extension

Abstract

In the research presented in the paper we try to find efficient methods for classification of MMPI profiles of patients with mental disorders. Each profile is described by a set of values of thirteen attributes (scales). Patients can be classified into twenty categories concerning nosological types. It is possible to improve classification accuracy by reduction or extension of the number of attributes with relation to the original data table. We test several techniques of reduction and extension. Experiments show that the proposed attribute extension approach improves classification accuracy, especially in the case of discretized data. © 2010 Springer-Verlag Berlin Heidelberg.