Sankhya: The Indian Journal of Statistics

2000, Volume 62, Series B, Pt. 2, pp. 233--248

NEW TRENDS IN MULTIVARIATE DIAGNOSIS

By

GENICHI TAGUCHI, American Supplier Institute, Livonia

and

J. RAJESH, Wayne State University, Detroit

SUMMARY. Taguchi Methods, also called Robust Design, have been successfully applied in many engineering applications to improve the performance of the product/process. These methods are proved to be extremely useful and cost effective. In this paper we will describe how Taguchi Methods are useful in multidimensional system applications. We use Mahalanobis distance (MD) for constructing a measurement scale for the multidimensional systems and the principles of Taguchi Methods for optimizing the system. Therefore, we refer to this procedure as the Mahalanobis-Taguchi-System (MTS). The measures and methods used in MTS are data analytic rather than usual probability based inference. In this paper, a detailed procedure of MTS is explained with two case studies.

AMS (1991) subject classification. 62H30, 62P30.

Key words and phrases. Mahalanobis distance (MD), Mahalanobis space (MS), Mahalanobis-Taguchi-System (MTS), Gram-Schmidt's orthogonalization process, medical diagnosis system, fire alarm sensor system, orthogonal arrays (OA's), signal-to-noise ratios (S/N ratios), direction of abnormals.

Full paper (PDF)