Analysis of Heavy Vehicle Crash Characteristics by Accounting for Heterogeneity – A Case Study of Freeways No. 1 & No. 3

Abstract

Due to heavy vehicles larger size and heavier weight, crashes involving these vehicles are extremely serious. In addition, since accident data is highly heterogeneous in nature, it needs to be processed to identify hidden relationships inherent in the data. Based on this, this study explores the severity characteristics of heavy vehicle crashes under the consideration of heterogeneity on freeway No. 1 & No. 3 in Taiwan. This study consists of three parts, including the suitable data proportion, heterogeneous data clustering and individual cluster crash severity models. The empirical study results show that protective equipment, lane number and speed camera are significantly and negatively associated with the occurrence of fatal crashes, indicating that they can reduce the probability of fatal crashes of heavy vehicles. While dark-unlighted, uphill, downhill and curve are significantly and positively associated with fatal crashes, suggesting that they can increase the probability of fatal crashes of heavy vehicles.