Analysis of Factors Contributing to the Severity of Large Truck Crashes

Abstract

Crashes that involved large trucks often result in immense human, economic, and social losses. To prevent and mitigate severe large truck crashes, factors contributing to the severity of these crashes need to be identified before appropriate countermeasures can be explored. In this research, we applied three tree-based machine learning (ML) techniques, i.e., random forest (RF), gradient boost decision tree (GBDT), and adaptive boosting (AdaBoost), to analyze the factors contributing to the severity of large truck crashes. Besides, a mixed logit model was developed as a baseline model to compare with the factors identified by the ML models. The analysis was performed based on the crash data collected from the Texas Crash Records Information System (CRIS) from 2011 to 2015. The results of this research demonstrated that the GBDT model outperforms other ML methods in terms of its prediction accuracy and its capability in identifying more contributing factors that were also identified by the mixed logit model as significant factors. Besides, the GBDT method can effectively identify both categorical and numerical factors, and the directions and magnitudes of the impacts of the factors identified by the GBDT model are all reasonable and explainable. Among the identified factors, driving under the influence of drugs, alcohol, and fatigue are the most important factors contributing to the severity of large truck crashes. In addition, the exists of curbs and medians and lanes and shoulders with sufficient width can prevent severe large truck crashes.

Rear-End Crash Risk Analysis considering Drivers’ Visual Perception and Traffic Flow Uncertainty: A Hierarchical Hybrid Bayesian Network Approach

“…e vehicle type (size) also significantly affects the probability of crash risk variations, as verified by 0.7% (0.8%) increase of the probability of "DECR" when "FVT � S (LVT � L)" was set as evidence. is was in line with the findings of Lee [30], Andersen et al [29], and Yoo and Green [33] indicating that the large leading vehicles would lead to a greater headway of the following vehicles and thus possibly result in less variations in crash risk as suggested by Ding et al [14,22]. e spatial frequency actually specified the effects of the perceptual markings on crash risk variations, and if it was set as "f s � L," the probability of "DECR" increased by 1.3% as compared with the original frequency.…”

“…According to the Highway Safety Manual [28] and to the nature of a surface transportation system, roadway crashes can be attributed to a combination of factors: the drivers (human factors), vehicles (vehicular factors), road infrastructure (roadway factors), and surrounding environment. Specifically, vehicle type and carfollowing mode (large/small vehicles following or being followed) are two main vehicular factors that have been widely verified to be influential in crashes and crash risk [14,22,[29][30][31][32][33]. Similarly, the roadway conditions, such as road type/grade, road geometric alignment, and road crosssection features, were extensively considered to account for crashes and crash risk (see eofilatos and Yannis [5]; Mannering et al [3]; and Papadimitriou et al [4] for systematic reviews).…”

A Systematic Review on the Role of Substance Consumption in Work-Related Road Traffic Crashes Reveals the Importance of Biopsychosocial Factors in Prevention

“…Based on the Texas Crash Records Information System (CRIS), Li and colleagues [ 10 ] analyzed 85,184 large truck RTC and found that 1.29% of the drivers (N = 1103) were under the influence of alcohol: 241 were classified as involved in serious RTC (incapacitating or fatal crashes), and 862 were classified as involved in non-serious RTC (all crash types excluding incapacitating or fatal crashes).…”

“…Considering all studies that used an objective measure and presented disaggregated data, it was possible to calculate the main descriptive statistics concerning the positivity rates for recreational drugs (summarized in Table 4 ). The positivity rates for recreational drugs among drivers involved in a RTC ranged from a low of 0.6% [ 10 ] to a high of 21.33% [ 14 ]; the weighted average (WA) and its confidence interval (95% CI) of the positivity rates to alcohol: WA= 0.84%; 95% CI (0.30–1.84%).…”

Employment of Artificial Intelligence Based on Routine Laboratory Results for the Early Diagnosis of Multiple Myeloma

“…This model outperforms other models as the direction of the negative gradient is followed in order to train the residuals of each iteration, which can avoid the over-fitting problem. Furthermore, the GBDT model has performed well for knowledge discovery in various fields (9,10). The current study is the first time that the artificial intelligence technology, including GBDT, has been used constructing a multiple myeloma early screening model based on a large amount of clinical conventional examination data.…”