Cyber-physical systems (CPS) interact with the physical and digital environment through embedded sensors and communication interfaces. Ever since sensors and data storage became highly available and affordable, CPS have been systematically deployed in the manufacturing industry. Consequently, huge amounts of data arise continuously in different systems and require integration and analysis. In this case study, sensor data from the window assembly of a German automotive manufacturing company is analyzed. By identifying factors that lead to deviations from required quality standards, production costs can ideally be decreased. The research benefit is a framework that covers data extraction, integration, preprocessing, analysis, and statistical inference. Different implemented supervised learning techniques as logistic regression and random forests enable the analysis of diverse datasets. Hence, the framework is applicable in many manufacturing scenarios where time series data arises at different locations. Further, the framework is designed such that manufacturing domain experts can easily be integrated into the analysis process.