Exploring truck driver-injury severity at intersections considering heterogeneity in latent classes: A case study of North Carolina
|نوع نگارش مقاله||
scopus – master journals – JCR
۴٫۲۷۶ در سال ۲۰۲۰
۲۶ در سال ۲۰۲۱
۰٫۹۰۱ در سال ۲۰۲۰
|شاخص Quartile (چارک)||
Q1 در سال ۲۰۲۰
خرید محصول توسط کلیه کارت های شتاب امکان پذیر است و بلافاصله پس از خرید، لینک دانلود محصول در اختیار شما قرار خواهد گرفت و هر گونه فروش در سایت های دیگر قابل پیگیری خواهد بود.
فهرست مطالب مقاله:
The fatal rate of truck-involved crashes is increasing and crashes become more severe than ۲۷ passenger vehicles in recent years. Much research has been dedicated to exploring the ۲۸ truck crash factors while scarce research focused on the intersection scenarios. This study ۲۹ investigates the factors that affect the severity level of truck-involved crashes at cross- and ۳۰
T-intersections. Due to the unobserved heterogeneity inherent in crash data, latent class ۳۱
|بخشی از متن مقاله:|
The truck-involved crashes have suffered more severe injury compared to passenger vehicles crashes in recent years.
53 According to U.S. Department of Transportation (USDOT) statistics, from 2016 to ۲۰۱۷, the number of large truck-
54 involved fatal crashes increased 9.6 % from 4251 to 4657, while passenger vehicle in fatal crashes has decreased 1.4%. Mean-
55 while, the fatal crash rate per ۱۰۰ million vehicle miles involving large trucks reached 1.42 and increased 5.2% from 2016 to
56 ۲۰۱۷, and it was ۱٫۳۸ times compared to the passenger vehicle (USDOT, 2019). In this case, a large number of research stud-
ies have been dedicated to investigations into the factors that impact the truck-involved crash severity. In considering that
58 different crash scenarios were a result of different influence factors, truck-involved crashes were specifically drawn into the
59 circumstances of human characteristics (Bernard and Mondy, 2016; Osman et al., 2018), roadway attributes (Ahmed et al.,
60 ۲۰۱۸; Wu et al., 2019a), location (Khorashadi et al., 2005; Uddin and Huynh, 2017), crash characteristics (Azimi et al., 2020),
61 vehicle characteristics (Uddin and Huynh, 2018), time (Anderson and Dong, 2017; Behnood and Mannering, 2019) and envi-
62 ronment (Uddin and Huynh, 2017). Intersections have more complex traffic circumstances and may cause more severe and
63 frequent crash injury compared to roadways (FHWA, 2004; Zhu and Srinivasan, 2011). Some studies mentioned intersection/
64 non-intersection (Wu et al., 2019b; Zhu and Srinivasan, 2011) or signal/non-signal control (Anderson and Dong, 2017; Chen
65 and Chen, ۲۰۱۱) as location or control type variables. However, scarce research on truck-involved crashes that specifically
66 considers the cross- and T-intersection scenarios has been conducted. Hence, it is important to explore the factors that con-
67 tribute to the truck-involved severity at cross- and T-intersections.
68 Even though many studies investigated the truck-involved crashes under specific conditions, there still remain many un-
69 observed factors that impact the crash severity and result in heterogeneity within the dataset. Research neglected the data’s
70 heterogeneity might generate wrong parameter estimations and conclusions (Song et al., 2020). Recently, many clustering
71 methods, such as k-means method (Mohamed et al., 2013), support vector method (Chen et al., 2016; Mokhtarimousavi,
72 ۲۰۱۹) and latent class analysis (Li and Fan, ۲۰۱۹; Liu and David, ۲۰۲۰; Mohamed et al., ۲۰۱۳; Sivasankaran and
73 Balasubramanian, 2020), were used to minimize the heterogeneity within the crash severity dataset. Mohamed et al.
74 (۲۰۱۳) combined multinomial logit and ordered probit model with k-means and latent class analysis method, and their
75 results confirmed that clustering the crash severity dataset into homogeneous clusters helps better identify factors that
76 would otherwise have been hidden without data segmentation. Since latent class analysis (LCA) is a model-based method
77 which could guarantee the homogeneity within the cluster based on statistical criterions, many traffic crash severity studies
78 recently implemented LCA for data segmentation (Fountas et al., 2018; Liu and David, ۲۰۲۰). Hence, this paper uses latent
79 class analysis to separate the dataset into groups which have the largest homogeneity within the same group.
80 For truck-involved crash severities, which are typically discrete in nature, a variety of ordered or unordered logit/probit
81 methods were implemented to conduct severity impact factors analysis, such as binary/multinomial logit model (Khorashadi
82 et al., 2005), Bayesian binary/multinomial logit model (Dong et al., 2017), mixed logit model (Anderson and Dong, ۲۰۱۷;
۸۳ Chen and Chen, ۲۰۱۱; Hou et al., ۲۰۱۹), ordered logit models (Osman et al., ۲۰۱۶), ordered probit model (Uddin and
84 Huynh, 2018), partial proportional odds model (Li and Fan, ۲۰۱۹) and random parameter ordered logit model (Azimi
85 et al., 2020).
86 For unordered model, Dong et al. (2017) used Bayesian multinomial logit and negative binomial model to investigate fac-
87 tors that affect large truck–involved crash frequency and severity based on ۲۰۰۶ to ۲۰۱۰ Tennessee data. Seat belt usage,
88 light condition, and terrain type were found to have significant effects only on the crash severity. In order to explore the
89 heterogeneity within the data, Anderson and Dong (۲۰۱۷) applied a mixed logit model to estimate heavy-vehicle crash fac-
90 tors based on ۲۰۰۴ to 2014 Minnesota data from HSIS database. Results showed that time-of-week needs to be considered
91 separately for safety analyses. Uddin and Huynh (2017) employed a mixed logit model to study the impacts of different light-
92 ing conditions on truck-involved crash severity in Ohio’s rural and urban areas during 2009 to 2013 from HSIS. The hetero-
93 geneous results indicated the importance of dividing data into different scenarios for more specific and accurate results.
94 The ordinal nature of crash severity (which usually increases from non-injury to fatal) violates the independence assump-
95 tion of the response variable for unordered model (Derr, ۲۰۱۳). Hence, many ordinary models were applied for better inves-
96 tigating the impact of the severity level. Chen et al. (2015) developed a hierarchical Bayesian random intercept model to
97 analyze the factors affecting rural truck-involved crash severity in New Mexico from 2010 to ۲۰۱۱٫ Results indicated the
98 existence of cross-level interaction effects between severity levels. Hassan et al. (2015) developed ordered probit and struc-
99 tural equation models to investigate truck crash severity’ factors and to study the impact of truck road based on the Abu
100 Dhabi data between 2007 and 2013. Results indicated that the likelihood of truck crashes involving fatalities was ۳۵% higher
101 on truck roads than that on mixed-vehicle roads. Azimi et al. (2020) constructed a random parameter ordered logit model to
102 detect potential sources of heterogeneity within large truck rollover crashes based on Florida’s 2007 to ۲۰۱۶ data. Results
103 showed significant variation within observations and the heterogeneous impacts on the severities. Osman et al. (2016) com-
104 pared ordered models with unordered models to analyze truck crash severity in work zones of Minnesota from HSIS. Results
105 showed that ordered logit has better model fitness than unordered models. By considering the ordinal feature of the crash
106 severities in the ordered model and investigating the heterogeneity characteristics of the crash data, this paper combines the
107 LCA with ordered logit model to further explore the factors that affect the truck-involved crash severity.
لطفا پیش از ارسال نظر، خلاصه قوانین زیر را مطالعه کنید: فارسی بنویسید و از کیبورد فارسی استفاده کنید. بهتر است از فضای خالی (Space) بیشازحدِ معمول، شکلک یا ایموجی استفاده نکنید و از کشیدن حروف یا کلمات با صفحهکلید بپرهیزید. نظرات خود را براساس تجربه و استفادهی عملی و با دقت به نکات فنی ارسال کنید؛ بدون تعصب به محصول خاص، مزایا و معایب را بازگو کنید و بهتر است از ارسال نظرات چندکلمهای خودداری کنید.