Pre analysis and clustering of uncertain data from manufacturing processes

With increasing complexity of manufacturing processes, the volume of data that has to be evaluated rises accordingly. The complexity and data volume make any kind of manual data analysis infeasable. At this point, data mining techniques become interesting. The application of current techniques is of complex nature because most of the data is captured by sensor measurement tools. Therefore, every measured value contains a specific error. In this paper, we propose an erroraware extension of the density-based algorithm DBSCAN. Furthermore, we discuss some quality measures that could be utilized for further interpretations of the determined clustering results. Additionally, we introduce the concept of pre-analysis during a necessary data integration step for the proposed algorithm. With this concept, the runtime of the error-aware clustering algorithm can be optimized and the integration of data mining in the overall software landscape can be promoted further .