Nonconformity root causes analysis through a pattern identification approach

Abstract

Controlling, maintaining, and improving quality is a central topic in manufacturing. Total Quality Management (TQM) provides several tools and techniques to deal with quality related topics, which are not always applicable. With the increased use of Information Technology (IT) in manufacturing there is a higher availability of data with great potential of further improvements. At the same time this results in higher requirements for data storage and processing with demanding, time consuming sessions for interpretation. Without suitable tools and techniques knowledge remains hidden in databases. This paper presents a methodology to help analyzing root causes of nonconformities (NCs) through a pattern identification approach. Hereby a methodology of Knowledge Discovery in Databases (KDD) is adapted and used as a quality tool. As the core element of the KDD methodology, the data mining, a well-known statistical measure from the field of economics—the Herfindahl–Hirschman Index (HHI)—is integrated. After presenting the theoretical background a new methodology is proposed and validated through an application case of the automotive industry. Results are obtained and presented in the form of patterns in matrices. They suggest that concentration indices may indicate possible root causes of NCs and invite for further investigations.