Pattern based decision tree analysis for risk detection in smart cities

Abstract

Increasing amounts of data on living environments and human interactions are becoming available. Their potential for valuable services improving the wellbeing of individuals is large and growing. This calls for an investigation of algorithms and system architectures that support possible use cases. In this paper we outline how pattern based decision tree analyses can be applied to the identification of risks caused by time-dependent effects from multiple influencing factors. For this purpose we apply the method to open data on car accidents and weather conditions. We also show how such systems can take advantage from up-to-date in-memory technology.