Title: Mahalanobis Taguchi System (MTS) and Mahalanobis Taguchi Gram-Schmidt (MTGS) methods as multivariate classification tools
Authors: Smarajit Bose; Rita SahaRay; Rohosen Bandyopadhyay
Addresses: Applied Statistics Division, Indian Statistical Institute, 203 B.T. Road, Kolkata-700 108, India ' Applied Statistics Division, Indian Statistical Institute, 203 B.T. Road, Kolkata-700 108, India ' Statistics Department, University of California, Davis, California, 95616, USA
Abstract: The Mahalanobis Taguchi System (MTS) and Mahalanobis Taguchi Gram-Schimdt (MTGS) methods were developed as diagnostic and predictive tools to separate between 'normal' and 'abnormal' data. The objective of these methods is to establish a measurement scale based on the 'normal' data so that the 'abnormal' data can be identified along with the degree of 'abnormality'. The goal of the present paper is to employ these methodologies as classification tools for multivariate data in general multi-class problems and compare the accuracy of the proposed tool with that of other existing multivariate classifiers using a variety of real life datasets.
Keywords: Mahalanobis-Taguchi system; MTS; Mahalanobis Taguchi Gram-Schimdt process; MTGS; Mahalanobis distance; Mahalanobis space; Taguchi methods; signal-to-noise ratio; S-N ratio; orthogonal arrays; classification tools; multivariate data.
International Journal of Industrial and Systems Engineering, 2014 Vol.16 No.1, pp.102 - 119
Available online: 28 Oct 2013 *Full-text access for editors Access for subscribers Purchase this article Comment on this article