Modeling of Low Visibility-Related Rural Single-Vehicle Crashes Considering Unobserved Heterogeneity and Spatial Correlation

Accident analysis and prevention are helpful to ensure the sustainable development of transportation. The aim this research was investigate factors associated with severity low-visibility-related rural single-vehicle crashes. Firstly, a latent class clustering model implemented partition whole-dataset into relatively homogeneous sub-dataset. Then, spatial random parameters logit established for each dataset capture unobserved heterogeneity correlation. Analysis conducted based on crash data (2014–2019) from 110 two-lane road segments. results show that proposed method is superior modeling approach accommodate Three variables—seatbelt not used, motorcycle, collision fixed object—have stable positive correlation severity. Motorcycle leads 12.8%, 23.8%, 12.6% increase in risk serious crashes whole-dataset, cluster 3, 4, respectively. In 2, caused by seatbelt used increased 5.5%, 0.1%, 30.6%, respectively, object 33.2%, 1.2%, 13.2%, can provide valuable information engineers policy makers develop targeted measures.