Assessing Urban Traffic Crash Risk and Its Impact on Mode Choice in Mumbai Using Safety Performance Functions

Abstract

Road traffic accidents continue to be a leading public safety concern in urban areas, and particularly for vulnerable road users including pedestrians and motorcyclists. While the increasing fatality and casualty counts urge a comprehensive approach to understand the factors contributing to crash risks, only a few studies in Mumbai focus on the dynamic interplay between the predictor variables of traffic volume, road condition, and types of road users in the prediction of crash risk. This study uses traffic data from Mumbai's road safety reports from 2019 to 2024 to develop a Negative Binomial Regression Model in order to predict crash risk. Safety Performance Functions are applied to estimate the influence of road features, traffic volume, location type, and weather conditions. The model is tested on real-world data, analyzing factors such as speed limits, road geometry, and user-specific risk factors. Some of the most significant factors that were identified in the model include the volume of traffic, geometric and location factors. The volume of traffic is an important factor, as the additional increase of 1,000 vehicles per day raises the risk of the accident by 0.23%. It has been predicted that 726 injuries have been caused by high-speed high-traffic peak-hour intersections, while 595 injuries have been caused by high-speed mid-block. The highest pedestrian crash risk was found to be in high-traffic intersections with 45.3%, compared to lower traffic intersections with only 15.2% in rural conditions. In terms of practical application, this model offers a means of identifying hazardous locations at a particular time that allows making effective measures to prevent accidents.