Abstract
The development of ride-hailing service has also led to an increase in the frequency of traffic accidents. The present study investigated the causes and accident severity for the ride-hailing accidents attributed to drivers-related factors in the context of China. From an online, self-reported survey of 1356 ride-hailing drivers across the country, we collected data about the drivers' demographic characteristics, working conditions, fatigue, risky driving behaviors and accident records between 2021 and 2023. We turned the data into insights through a two-step approach: using a Random Forest (RF) model to identify the most significant factors influencing accident severity, followed by building a Bayesian Network (BN) model to analyze the relationships between the identified factors and accident severity. With 16 top factors according to the RF model, results from the BN model showed that the main risk factors differ between different levels of accident severity. Among all the factors, nine proved to be directly related to accident severity, mostly involving drowsiness, using smartphones in inappropriate situations and risky driving behaviors; the drivers' demographic and working conditions otherwise influence accident severity in indirect ways. The findings from this study are useful for proposing more targeted policies to mitigate the accident severity among ride-hailing drivers.