Predictive Analytics for Traffic Safety
Assessing the applicability of predictive policing and predictive analytics for improving traffic safety
The motivation behind predictive policing is to predict crimes before they occur. The PreASiSt project assessed whether general concepts of predictive policing can be adopted and applied to the field of traffic safety.
Models should help to understand and capture conditions that could trigger an accident. Forecasts of these variables can be used to predict and evaluate the risk of accidents. Based on these risks, the police can plan and take measures to increase traffic safety in these areas. In order to identify relevant parameters, this project approached the problem from a theoretical and data driven view point, using the Cross Industry Standard Process for Data Mining (CRISP). The city of Bremen provided a detailed history of traffic accidents and other relevant data sets and their local police contributed expert knowledge.
52°North focused on data analytics. We investigated both the suitability of available data sources for predictive data analysis and the use of machine learning methods for creating accident risk prediction models. Data understanding; however, can only be achieved by understanding the underlying processes. Thus, we worked in close collaboration with the Institute of Traffic and Engineering Psychology of the German Police University (DHPOL) and the local experts from the Bremen Police.
The project came to a close in 2020. For our part, our team concluded the data analysis. This focused on the development of accident density maps based on external conditions, such as the day of the week or the traffic conditions on the surrounding highways, and their impact on the inner city traffic. Furthermore, we employed machine learning approaches to extract relevant parameters affecting the occurrence of accidents.