Abstract—High competitive pressure in the global manufac- turing industry makes efficient, effective and continuously improved manufacturing processes a critical success factor. Yet, existing analytics in manufacturing, e. g., provided by Manufacturing Execution Systems, are coined by major short- comings considerably limiting continuous process improvement. In particular, they do not make use of data mining to identify hidden patterns in manufacturing-related data. In this article, we present indication-based and pattern-based manufacturing process optimization as novel data mining approaches provided by the Advanced Manufacturing Analytics Platform. We demonstrate their usefulness through use cases and depict suitable data mining techniques as well as implementation details. Index Terms—Analytics, Data Mining, Decision Support, Process Optimization I. INTRODUCTION A. Motivation Globalization, shorter product lifecycles and rapidly changing customer needs lead to high competitive pressure in the manufacturing industry. Apart from product quality and product variety, flexibility, short lead times and a high adherence to delivery dates have become essential success factors [1]. Thus, efficient, effective and continuously opti- mized manufacturing processes are central prerequisite to perform successfully on the market [2]. Looking at other industry sectors, Business Intelligence (BI) technology is successfully applied for the optimization of workflow-based business processes, esp. in the service industry [3], [4]. This emphasizes the potential of using comprehensive analytics to improve business activities. Regarding BI approaches in manufacturing, there are mainly two types, wide-spread in industry practice: On the one hand, pre-packaged dashboard applications based on Manuscript received March 14, 2012; revised April 12, 2012. The au- thors would like to thank the German Research Foundation (DFG) for financial support of this project as part of the Graduate School of Excel- lence advanced Manufacturing Engineering (GSaME) at the University of Stuttgart, Germany. C. Gröger is with the Institute of Parallel and Distributed Systems (IPVS), University of Stuttgart, 70569 Stuttgart, Germany. He is a member of the Graduate School of Excellence advanced Manufacturing Engineer- ing (GSaME) at the University of Stuttgart (phone: +49 711 685-88242; fax: +49 711 685-88424; e-mail: christoph.groeger@ipvs.uni-stuttgart.de). F. Niedermann and B. Mitschang are with the Institute of Parallel and Distributed Systems (IPVS), University of Stuttgart, 70569 Stuttgart, Germany (e-mail: florian.niedermann@ipvs.uni-stuttgart.de, bern- hard.mitschang@ipvs.uni-stuttgart.de). metrics visualization and basic reporting, typically part of Manufacturing Execution Systems (MES) [5]; on the other hand, custom BI applications that mainly focus on spread- sheet-based Online Analytical Processing (OLAP) [6]. The- se existing BI approaches are coined by the following major shortcomings, considerably limiting continuous process improvement:  Being based on isolated data extracts, they do not adopt a holistic view integrating operational and process data, e. g., from MES and Enterprise Resource Planning (ERP) Systems.  They focus on OLAP-like analysis and classical report- ing and do not employ advanced analytics techniques, esp. data mining, to extract knowledge from data.  They only provide limited means for sharing and com- bination of analysis results, for example in different sub processes of Manufacturing Process Management.  They offer no guidance for transforming analysis results into concrete process modifications – leaving this step entirely up to the subjective judgement and skills of the process analyst. Eliminating these insufficiencies is the key motivation of the Advanced Manufacturing Analytics (AdMA) Platform, which is being developed as part of our overall work. In this article, we focus on indication-based and pattern-based op- timization as novel concepts for process-centric data mining in manufacturing provided by the AdMA Platform. The remainder is organized as follows: First, we introduce the AdMA Platform and characterize existing data mining approaches in manufacturing in Section 2. Next, we present Indication-based and Pattern-based Manufacturing Optimi- zation in Section 3. Section 4 details the former and defines corresponding uses cases. In addition, adequate data mining techniques for a selected use case are discussed and the prototypical implementation as well as a first proof of con- cept is presented. We conclude in Section 5 and point out future work. B. The Advanced Manufacturing Analytics Platform The Advanced Manufacturing Analytics Platform [7] is an integrated BI platform for holistic data-driven manufacturing process optimization. It is based on a transfer of concepts of the Deep Business Optimization Platform [8], [3], [9] to the area of manufacturing. Its conceptual architecture consists of three integrated layers sketched in Fig. 1. Data Mining-driven Manufacturing Process Optimization Christoph Gröger, Florian Niedermann, and Bernhard Mitschang Proceedings of the World Congress on Engineering 2012 Vol III WCE 2012, July 4 - 6, 2012, London, U.K. ISBN: 978-988-19252-2-0 ISSN: 2078-0958 (Print); ISSN: 2078-0966 (Online) WCE 2012