Bayesian random effect models incorporating real-time weather and traffic data to investigate mountainous freeway hazardous factors

► This study includes real-time weather data and real-time traffic data to analyze crash frequency on a mountainous freeway. ► Three models are compared - the Poisson model, the uncorrelated random effect model and the correlated random effect model. ► Two scenarios have been considered - the seasonal model and the crash type (single vehicle crashes and multi-vehicle crashes) model. ► Bayesian inference method have been employed to analyze the model, and the results indicate that real-time weather variables and traffic flow variables play a key role in crash occurrences and the correlated random effect model outperformed the others.