| Abstrakt: |
Crash severity outcomes are random and are influenced by various interactions between factors like road geometry, roadside hazards, pavement characteristics and crash-related factors. This study aims to assess the effect of pavement and shoulder conditions, types and markings, lighting and weather conditions and crash factors like vehicles involved, etc. on the severity outcomes of crashes for urban roads having heterogeneous traffic flow using structural equation modelling (SEM) and nonparametric multivariate adaptive regression splines (MARS). The study used 974 recorded crashes that occurred in Patna, India, during years 2015–2017. Crash severity outcomes were considered at four levels, namely property damage only, minor injury, major injury and fatal. It was observed that MARS model has better capability of modelling crash severity outcomes compared to SEM model when influence of many attributes is analysed with limited data size. However, SEM model could capture the complex interaction between latent and measured variables for modelling crash severity outcomes. It was observed that vehicle type, accident type, lighting condition and pavement condition are important parameters influencing crash severity outcomes on low speed urban roads. |
Full text from SpringerLink