Modeling over speeding behavior of vehicles using a random parameter negative binomial approach: A case study of Mumbai, India

Abstract

Over speeding is one of the most common factors contributing to road accidents leading to traffic-related injuries and deaths. Understanding determinants of over speeding behavior is an important step toward enhancing road safety and developing robust intervention strategies. Moreover, there is a need to extend this knowledge with respect to developing countries, where there is a significant variation in the operating speeds of the vehicles owing to the heterogeneous traffic conditions. The present study captured the speeding behavior of the vehicles in Mumbai, India, during the years 2017 and 2018 using roadside observational surveys along with the spatial and temporal attributes of the selected locations. This study investigates the combined effect of temporal and spatial attributes on the over speeding frequency. A random parameter negative binomial model is developed utilizing the observational data to explain the over speeding behavior of the vehicles at the selected count locations. The developed model accounts for unobserved heterogeneity across the over speeding count locations based on temporal, road geometric, and built environment attributes. The study results confirm the presence of heterogeneity in the over speeding behavior of the vehicles at the count locations. In terms of policy implications, results suggest that temporary speed calming strategies could be adopted during the non-peak hours and on weekends. Similarly, deploying speed humps, rumble strips, or adopting law enforcement strategies can circumvent over speeding behavior on arterial roads. Lastly, developing well-connected roads with frequent presence of intersections and traffic signals can be another strategy applicable for such behavior.