Modeling the Effects of Weather and Traffic on the Risk of Secondary Incidents

Secondary incidents have increasingly been recognized as a major source of freeway incidents as their occurrence causes significant additional traffic delay. In this paper, past research is extended in two directions; first, real-time traffic and weather data is incorporated to examine their effects on the likelihood of secondary incidents. Second, neural network classifiers and popular statistical classification approaches are developed and findings and predictability compared. Results suggest that changes in speed and volume, number of blocked lanes and number of vehicles involved significantly influence the probability of having a secondary incident, whereas changes in rainfall intensity are less influential to secondary incident likelihood.