abstract
- © 2019 The Authors. Published by Elsevier B.V.In today's manufacturing environment, most mature organizations generate only a few Defects Per Million of Opportunities (DPMO). Detecting these defects is one of the main goals of Process Monitoring for Quality (PMQ). The Support Vector Machine (SVM) algorithm has been success-fully applied to perform this task. PMQ is a big data driven quality philosophy founded on big models, which is a predictive modeling paradigm aimed at developing a manufacturing-functional model. Due to the combinatorial explosion associated to big data-based analyzes, many models need to be created and tested. Model selection becomes a critical and integral aspect of scientific data analysis that leads to the development of a functional model. A three-objective optimization model selection criterion (3D - SVM) is introduced for analyzing highly/ultra unbalanced data structures. It uses three competing attributes - prediction, separability, complexity - to project each candidate model into a 3D-space to select the final model. The 3D - SVM criterion is enabled by a new separability index introduced here, and is a step forward in the application of machine learning initiatives for rare event detection.