Identifying and profiling hazardous locations by means of quantitative techniques

Abstract

In this research, we had two main research objectives. First, we wanted to investigate how hazardous accident locations are currently identified and ranked in Flanders (the Flemish speaking community of Belgium). More specifically, the objective was to perform a sensitivity analysis, based on the same data used to select and rank the 1,014 accident sites that are currently considered as dangerous by the Flemish government, to investigate the strengths and weaknesses of this approach to rank and select dangerous accident sites. Second, in order to develop effective countermeasures to reduce the number of accidents at dangerous locations, one should properly and systematically relate accident frequency and severity to a large number of variables. Accordingly, the second objective of this research was to investigate how data mining and statistical techniques can be implemented to identify and pro¯le dangerous accident locations in terms of accident related data and location characteristics. Additionally, we aimed to develop new models, using these techniques, which will provide new insights into the criteria that aim to explain to probability of the occurrence of a tra±c accident. Finally, we were also interested in investigating whether the dangerous accident sites in Flanders tend to migrate over time using accident data from different time periods.