Disaggregated crash prediction models for different crash types using joint probability model

Road safety affects health and development worldwide, thus it is essential to examine the factors that influence crashes and injuries. As the relationships between crashes, the severity of injuries, and possible risk factors could vary depending on the type of collision, we attempt to develop disaggregated prediction models for different crash types in terms of number of vehicles involved in the crash (i.e., single- and multi-vehicle crashes). As disaggregated speed data are available from the probe vehicles equipped with global positioning system (GPS) devices, it is possible to determine how average speed and its dispersion affect crashes and injries. We also examine risk factors, including geometric design, weather conditions, and temporal distribution. A comprehensive database is thus established with information on road crashes and other possible risk factors for selected road segments in Hong Kong. In particular, a joint probability model is applied to analyze crash occurrence and severity simultaneously under an integrated modeling framework. As a result, speed plays an important role in road safety performance. Furthermore, rainfall and several geometric factors are also significantly associated with crash occurrence or severity.