Development of High-dimensional Data Analysis Procedures in Mahalanobis-Taguchi System

Abstract

The Mahalanobis-Taguchi (MT) system, which is representative of the Taguchi method, plays an important role in various fields centered on the manufacturing industry. Within these fields, the MT method proposed at the earliest stage has been applied to various data as a practical anomaly detection procedure, and there are abundant application cases. However, it is clear that the con- ventional MT method cannot always perform appropriate analyses. Therefore, due to the practical demand for the MT system, we have proposed novel anomaly detection procedures to appropriately analyze real problems. In this paper, we comprehensively report our research results on the MT system, especially for high-dimensional data.