Data quality in the automotive industry
Measurement and analysis of data quality in time series data from production processes
Measurement and analysis of data quality in time series data from production processes
Prof. Dr. Mathias Klier
Andreas Obermeier
Torben Widmann
Prof. Dr. Mathias Klier
+49 (0) 7 31 50-3 23 12
mathias.klier(at)uni-ulm.de
In the course of digitalization, organizations have access to very extensive and ever-growing volumes of data (keyword: "big data"). The volume of data will increase tenfold between 2016 and 2025, creating a wide range of opportunities for companies. For example, the targeted and well-founded analysis and use of this data enables improved, data-driven decision support and process management. Particularly since the start of Industry 4.0, the importance of data in the production environment and the associated acquisition of insights as well as the generation of efficiency gains has been steadily increasing. Ensuring sufficient data quality is an essential basis for the profitable use of data. If this is not sufficiently given, erroneous analysis results and wrong decisions result, which cause more harm than good ("garbage in, garbage out"). However, empirical evidence shows that the data used by companies is often characterized by low data quality. The main reasons for this are insufficient sensitivity to data quality and data quality problems, a lack of transparency of the data quality level (data quality measurement), and hardly any systematic quality improvement and control.
Against this background, there is also a need for our practice partner from the automotive industry to create the basis for a methodical data quality management. In the production environment, data from the various production technologies and controls in particular represent a central asset and an important source of Big Data. In a joint pilot project, an analysis is to be carried out for a standard technology in production in order to assess and, if necessary, improve the status of data quality. For this purpose, meaningful KPIs (Key Performance Indicators) will be defined for the quality of the time series data of the production process, suitable metrics for measurement will be developed and a quantification of the data quality will be derived. The goal is to gain knowledge about the current data quality through the specially developed methods and procedures and, if necessary, to derive improvements in decisions for action and to reduce waste. Furthermore, the results of this project are to be extended and transferred to many other fields of application in the future.
Cooperation partners: Premium manufacturers from the German automotive industry
Project period: December 2020 - April 2021
This pilot project aims to assess the quality of the large amount of data ("Big Data") collected within the production environments of the automotive industry, i.e. to quantify it with meaningful and interpretable metrics. These metrics add to the body of scientific methods in the area of data quality with which the Institute of Business Analytics is significantly advancing this branch of research. The improved assessment of data quality in the production environment makes it possible to establish a comprehensive data quality management and to proactively reduce losses, e.g. through reduced wastage. The cooperation and transfer in this project thus creates a direct added value for science and practice.