In an environment of steadily increasing competitiveness, data-driven product development offers a great opportunity for automotive manufacturers. Vehicle usage data can be used, for example, to optimize technical components for their real-life stress or to develop new customer functions based on real user behavior. With the increasing interconnectedness of modern vehicles, usage data in the form of vehicle bus signals can be used for data-driven development. However, the use of this data often poses a challenge because it was designed for internal communication of the vehicle rather than for later analysis. Accordingly, this paper presents how data mining (DM) methods can be used to extract customer behavior from this data. The central idea of the presented methods is the derivation and enrichment of the vehicle bus signals with metadata of the vehicle usage, for example but not exclusively by applying statistical methods like machine learning. The method of metadata enrichment is embedded in an adapted version of the Cross Industry Standard Process for Data Mining (CRISP-DM). The developed methods are presented based on two concrete application examples and finally a general recommendation for action for data-driven product development in the automotive sector is derived.
Session:
DATA SCIENCE
|
| 09:30 - 10:00