A STATISTICAL CONCEPT IN DETERMINATION OF THRESHOLD VALUE FOR FUTURE DIAGNOSIS IN MTS: AN ALTERNATIVE TO TAGUCHI?S LOSS FUNCTION APPROACH

Mahalanobis-Taguchi System (MTS) is a pattern information technology, which has been used in different diagnostic applications to make quantitative decisions by constructing a multivariate measurement scale using data analytic methods, without any assumption regarding statistical distribution. It uses a threshold value computed through quadratic loss function approach for the future diagnosis. However, the procedure for determining the threshold is lacking statistical explanation and primarily dependent on the domain expertise, if available, from the respective field. This study makes an alternate attempt to determine a threshold value using the property of multivariate statistical distributions related to Mahalanobis D2, used for MTS, and the critical values of the distributions for different levels of significance. The findings of the study can be used to eliminate the subjectivity of MTS in compliance to its data analytic philosophy.