Driver Celeration Behavior and the Prediction of Traffic Accidents

A study was undertaken to investigate whether driver celeration (overall mean speed change) behavior can predict traffic accident involvement. Also, to test whether acceleration, deceleration or the combined celeration measure was the better predictor. Bus driver celeration behavior was measured repeatedly in real traffic, driving en route, and correlated with accidents for which the drivers were deemed at least partly responsible. Correlations around .20 were found in several samples between celeration behavior and culpable accidents for a 2-year period. The results show that although celeration behavior is only semi-stable over time, it predicts with some accuracy individual accident involvement over 2 years. The predictive power of acceleration and deceleration was slightly lower than the combined measure, in accordance with theory. The correlations found were strong enough to warrant the use of celeration behavior as a predictive variable for transportation companies in their safety work.     

Driver celeration behavior and the prediction of traffic accidents.

A study was undertaken to investigate whether driver celeration (overall mean speed change) behavior can predict traffic accident involvement. Also, to test whether acceleration, deceleration or the combined celeration measure was the better predictor. Bus driver celeration behavior was measured repeatedly in real traffic, driving en route, and correlated with accidents for which the drivers were deemed at least partly responsible. Correlations around .20 were found in several samples between celeration behavior and culpable accidents for a 2-year period. The results show that although celeration behavior is only semi-stable over time, it predicts with some accuracy individual accident involvement over 2 years. The predictive power of acceleration and deceleration was slightly lower than the combined measure, in accordance with theory. The correlations found were strong enough to warrant the use of celeration behavior as a predictive variable for transportation companies in their safety work.     

Effects of passengers on bus driver celeration behavior and incident prediction

PROBLEM: Driver celeration (speed change) behavior of bus drivers has previously been found to predict their traffic incident involvement, but it has also been ascertained that the level of celeration is influenced by the number of passengers carried as well as other traffic density variables. This means that the individual level of celeration is not as well estimated as could be the case. Another hypothesized influence of the number of passengers is that of differential quality of measurements, where high passenger density circumstances are supposed to yield better estimates of the individual driver component of celeration behavior. METHOD: Comparisons were made between different variants of the celeration as predictor of traffic incidents of bus drivers. The number of bus passengers was held constant, and cases identified by their number of passengers per kilometer during measurement were excluded (in 12 samples of repeated measurements). RESULTS: After holding passengers constant, the correlations between celeration behavior and incident record increased very slightly. Also, the selective prediction of incident record of those drivers who had had many passengers when measured increased the correlations even more. CONCLUSIONS: The influence of traffic density variables like the number of passengers have little direct influence on the predictive power of celeration behavior, despite the impact upon absolute celeration level. Selective prediction on the other hand increased correlations substantially. This unusual effect was probably due to how the individual propensity for high or low celeration driving was affected by the number of stops made and general traffic density; differences between drivers in this respect were probably enhanced by the denser traffic, thus creating a better estimate of the theoretical celeration behavior parameter C. The new concept of selective prediction was discussed in terms of making estimates of the systematic differences in quality of the individual driver data.     

Aggregation of driver celeration behavior data: Effects on stability and accident prediction

Predictions about effects of aggregating driver celeration data were tested in a set of data where bus drivers’ behavior had been measured repeatedly over three years in a city environment. For drivers with many measurements, this data was correlated with the drivers’ accident record at various levels of aggregation over measurements. A single measurement (one sample) was seldom a significant predictor, but for each drive added to a mean, the variation explained in accident record was increased by about 1%. Also, correlations between measurements increased when these were aggregated, and the association with number of passengers (a proxy for traffic density) decreased somewhat, all as predicted. These results show that although driver celeration behavior is only semi-stable across time and environments, aggregating measurements increases both stability and predictive power versus accidents considerably. The celeration variable is therefore promising as a tool for identifying dangerous drivers, if these can be measured repeatedly, or, even better, continuously.     

The relation of non-culpable traffic incidents to bus drivers' celeration behavior

PROBLEM: The driver celeration behavior theory predicts that celerations are associated with incidents for which the driver has some responsibility in causing, but not other incidents. METHOD: The hypothesis was tested in 25 samples of repeated measurements of bus drivers' celeration behavior against their incidents for two years. RESULTS: The results confirmed the prediction; in 18 samples, the correlation for culpable incidents only was higher than for all incidents, despite the higher means of the latter. Non-culpable incidents had correlations close to zero with celeration. DISCUSSION: It was pointed out that most individual crash prediction studies have not made this differentiation, and thus probably yielded underestimates of the associations sought, although the effect is not strong, due to non-culpable accident involvements being few (less than a third of the total). The methods for correct identification of culpable incident involvements were discussed.     

Long-Term Prediction of Traffic Accident Record From Bus Driver Celeration Behavior

Driver celeration (speed change) behavior of bus drivers measured a number of times was used to predict their culpable accidents over increasing time periods. It was found that predictive power was considerable (>.30 correlation) over 5 years of time with aggregated celeration (mean of repeated measurements) as independent variables, and there were also indications that power reached even further, although too low Ns made these results unreliable. Similarly, there were indications of even stronger correlations with increased aggregation of celeration values. The results were discussed in terms of the methodology needed to bring out such results, and the stability of accident-causing behavior over time.     

Long-term prediction of traffic accident record from bus driver celeration behavior.

Driver celeration (speed change) behavior of bus drivers measured a number of times was used to predict their culpable accidents over increasing time periods. It was found that predictive power was considerable (>.30 correlation) over 5 years of time with aggregated celeration (mean of repeated measurements) as independent variables, and there were also indications that power reached even further, although too low Ns made these results unreliable. Similarly, there were indications of even stronger correlations with increased aggregation of celeration values. The results were discussed in terms of the methodology needed to bring out such results, and the stability of accident-causing behavior over time.     

Random parameter models for accident prediction on two-lane undivided highways in India

Introduction

Generalized linear modeling (GLM), with the assumption of Poisson or negative binomial error structure, has been widely employed in road accident modeling. A number of explanatory variables related to traffic, road geometry, and environment that contribute to accident occurrence have been identified and accident prediction models have been proposed. The accident prediction models reported in literature largely employ the fixed parameter modeling approach, where the magnitude of influence of an explanatory variable is considered to be fixed for any observation in the population. Similar models have been proposed for Indian highways too, which include additional variables representing traffic composition. The mixed traffic on Indian highways comes with a lot of variability within, ranging from difference in vehicle types to variability in driver behavior. This could result in variability in the effect of explanatory variables on accidents across locations. Random parameter models, which can capture some of such variability, are expected to be more appropriate for the Indian situation.

Method

The present study is an attempt to employ random parameter modeling for accident prediction on two-lane undivided rural highways in India. Three years of accident history, from nearly 200 km of highway segments, is used to calibrate and validate the models.

Results

The results of the analysis suggest that the model coefficients for traffic volume, proportion of cars, motorized two-wheelers and trucks in traffic, and driveway density and horizontal and vertical curvatures are randomly distributed across locations.

Conclusions

The paper is concluded with a discussion on modeling results and the limitations of the present study.     

Data mining of tree-based models to analyze freeway accident frequency

INTRODUCTION: Statistical models, such as Poisson or negative binomial regression models, have been employed to analyze vehicle accident frequency for many years. However, these models have their own model assumptions and pre-defined underlying relationship between dependent and independent variables. If these assumptions are violated, the model could lead to erroneous estimation of accident likelihood. Classification and Regression Tree (CART), one of the most widely applied data mining techniques, has been commonly employed in business administration, industry, and engineering. CART does not require any pre-defined underlying relationship between target (dependent) variable and predictors (independent variables) and has been shown to be a powerful tool, particularly for dealing with prediction and classification problems. METHOD: This study collected the 2001-2002 accident data of National Freeway 1 in Taiwan. A CART model and a negative binomial regression model were developed to establish the empirical relationship between traffic accidents and highway geometric variables, traffic characteristics, and environmental factors. RESULTS: The CART findings indicated that the average daily traffic volume and precipitation variables were the key determinants for freeway accident frequencies. By comparing the prediction performance between the CART and the negative binomial regression models, this study demonstrates that CART is a good alternative method for analyzing freeway accident frequencies. IMPACT ON INDUSTRY: By comparing the prediction performance between the CART and the negative binomial regression models, this study demonstrates that CART is a good alternative method for analyzing freeway accident frequencies.     

Analyses of rear-end crashes based on classification tree models

OBJECTIVE: Signalized intersections are accident-prone areas especially for rear-end crashes due to the fact that the diversity of the braking behaviors of drivers increases during the signal change. The objective of this article is to improve knowledge of the relationship between rear-end crashes occurring at signalized intersections and a series of potential traffic risk factors classified by driver characteristics, environments, and vehicle types. METHODS: Based on the 2001 Florida crash database, the classification tree method and Quasi-induced exposure concept were used to perform the statistical analysis. Two binary classification tree models were developed in this study. One was used for the crash comparison between rear-end and non-rear-end to identify those specific trends of the rear-end crashes. The other was constructed for the comparison between striking vehicles/drivers (at-fault) and struck vehicles/drivers (not-at-fault) to find more complex crash pattern associated with the traffic attributes of driver, vehicle, and environment. RESULTS: The modeling results showed that the rear-end crashes are over-presented in the higher speed limits (45-55 mph); the rear-end crash propensity for daytime is apparently larger than nighttime; and the reduction of braking capacity due to wet and slippery road surface conditions would definitely contribute to rear-end crashes, especially at intersections with higher speed limits. The tree model segmented drivers into four homogeneous age groups: < 21 years, 21-31 years, 32-75 years, and > 75 years. The youngest driver group shows the largest crash propensity; in the 21-31 age group, the male drivers are over-involved in rear-end crashes under adverse weather conditions and the 32-75 years drivers driving large size vehicles have a larger crash propensity compared to those driving passenger vehicles. CONCLUSIONS: Combined with the quasi-induced exposure concept, the classification tree method is a proper statistical tool for traffic-safety analysis to investigate crash propensity. Compared to the logistic regression models, tree models have advantages for handling continuous independent variables and easily explaining the complex interaction effect with more than two independent variables. This research recommended that at signalized intersections with higher speed limits, reducing the speed limit to 40 mph efficiently contribute to a lower accident rate. Drivers involved in alcohol use may increase not only rear-end crash risk but also the driver injury severity. Education and enforcement countermeasures should focus on the driver group younger than 21 years. Further studies are suggested to compare crash risk distributions of the driver age for other main crash types to seek corresponding traffic countermeasures.     

In-vehicle workload assessment: effects of traffic situations and cellular telephone use

PROBLEM: Assessment of drivers' on-road workload is an important traffic safety consideration. This study was conducted to examine the effects of cellular phone communication on driving performance, with particular emphasis on variations in task demand in different traffic situations. METHOD: Twelve participants were asked to drive on urban roads and motorways with or without concomitant mathematical-addition tests relayed via cellular phone. Measurements included task and driving performance, physiological responses, and compensatory behavior. RESULTS: Analysis of task performance revealed that mean response time was markedly increased (11.9%) for driving on urban roads compared to motorways. The mean driving speed only decreased 5.8% in the presence of phone tasks in comparison to normal driving without distractions. In addition, overall physiological workload increased through compensatory behavior in response to the phone tasks. CONCLUSIONS: Driving with phone use in different traffic environments induced measurable variations in driver workload. IMPACT ON INDUSTRY: When faced with heavy traffic, a greater safety margin is typically adopted, with more lowered driving speed and restricted phone use, and it can be assumed that there is a general trade-off between tasks to preserve driving safety.     

基于车速的交通事故贝叶斯预测

为了降低交通事故的发生率 ,提高道路交通安全水平 ,提出了基于车速的贝叶斯预测方法来检测和预测交通事故。首先对车速与交通事故之间的关系进行分析。在分析的基础上 ,以车速为衡量对象 ,提出贝叶斯预测方法。通过使用车速观测数据 ,应用 χ2 检验 ,确定是否为异常数据 ;并通过最小风险的贝叶斯预测 ,确定该异常是否会导致交通事故。最后 ,绘出利用该贝叶斯预测方法进行交通事故预测的流程图     

Behavior and lifestyle characteristics of male Kuwaiti drivers

Introduction: The high traffic accident risk among young drivers is a well-known and well-documented fact in most countries. Lifestyle has proven to affect driving behavior as well as accident risk. This study covers the lifestyle component of the problems related to young male Kuwaiti drivers’ accident risk. Methods: The purpose of the study is to measure the relationship between lifestyle and accident risk. Lifestyle is measured through a questionnaire, where 302 male Kuwaiti drivers (mean age = 28 years; range 25-35 years) answer 39 questions related to behavioral and social factors, road conditions, police enforcement, and life satisfaction. They also report their involvement in accidents and traffic violations. Results: The questionnaire's validity and reliability (Cronbach's alpha = 0.7) were achieved. Principal component analysis reduced the 39 items on the questionnaire to 5 factors. Inadequate police enforcement is strongly correlated (r = 0.862) to accident risk and traffic violations and is thus considered the best predictor of traffic accidents in Kuwait. Impact on Industry: As driving-related incidents (on-the-job and off-the-job) are a significant source of fatalities and lost-work-days, the study points to the importance of considering cultural factors in the design of comprehensive safety programs for industry.     

我国高速公路交通事故趋势分析

对高速公路交通事故的主要原因、事故类型和发展趋势作了分析和预测 ,拟合出每百公里交通事故起数的预测公式 ,指出应加强对超速行车、通车初期和恶劣天气时的交通控制。分析结论是 :高速公路交通事故总体上呈下降趋势 ;事故的绝对数将继续增长 ,但每百公里事故起数将逐年平缓下降 ;在高速公路交通事故中 ,单方事故居多 ,超速行驶是引发事故的主要原因 ,追尾是事故的主要形态 ;在高速公路通车初期和恶劣气候条件下 ,事故呈上升趋势     

Applying machine learning approaches to analyze the vulnerable road-users' crashes at statewide traffic analysis zones

Introduction: In this paper, we present machine learning techniques to analyze pedestrian and bicycle crash by developing macro-level crash prediction models. Methods: We collected the 2010–2012 Statewide Traffic Analysis Zone (STAZ) level crash data and developed rigorous machine learning approach (i.e., decision tree regression (DTR) models) for both pedestrian and bicycle crash counts. To our knowledge, this is the first application of DTR models in the burgeoning macro-level traffic safety literature. Results: The DTR models uncovered the most significant predictor variables for both response variables (pedestrian and bicycle crash counts) in terms of three broad categories: traffic, roadway, and socio-demographic characteristics. Additionally, spatial predictor variables of neighboring STAZs were considered along with the targeted STAZ in both DTR models. The DTR model considering spatial predictor variables (spatial DTR model) were compared without considering spatial predictor variables (aspatial DTR model) and the model comparison results discovered that the prediction accuracy of the spatial DTR model performed better than the aspatial DTR model. Finally, the current research effort contributed to the safety literature by applying some ensemble techniques (i.e. bagging, random forest, and gradient boosting) in order to improve the prediction accuracy of the DTR models (weak learner) for macro-level crash count. The study revealed that all the ensemble techniques performed slightly better than the DTR model and the gradient boosting technique outperformed other competing ensemble techniques in macro-level crash prediction models.     

基于Hankel-DMD的城市交通事故风险时空预测*

为解决城市交通事故风险时空分布预测任务中时空关联性捕捉困难的问题,提出基于动态模态分解（DMD）的城市交通事故分析时空预测模型,模型利用总最小二乘法去除交通事故数据中的噪声,应用结合Hankel矩阵的动态模态分解模型（Hankel-DMD）捕捉交通事故风险的时空关联性,对交通事故风险的时空分布进行预测。研究结果表明:DMD框架能够为高维预测任务提供低秩解决方案,从高维数据中捕捉时空关联性;Hankel-DMD模型在预测评价指标平均绝对误差和均方根误差方面的表现明显优于统计学及机器学习等方法;Hankel-DMD模型产生的动态模态和特征值,对事故风险系统的时空动态特征具有一定的可解释性,同时验证Hankel-DMD模型的适用性。     

Factors associated with driving under the influence of alcohol

Objective: This study examined the risk factors of driving under the influence of alcohol (DUI) among drivers of specific vehicle categories (DSC). On the basis of this research, the variables related to DUI and involvement in traffic crashes were defined. The analysis was conducted for car drivers, bicyclists, motorcyclists, bus drivers, and truck drivers.

Method: The research sample included drivers involved in traffic crashes on the territory of Serbia in 2016 (60,666). Two types of analyses were conducted in this study. Logistic regression established the correlation between DUI and DSC and the The Technique for Order of Preference by Similarity to Ideal Solution (Multi-criteria decision making) method was applied to consider the scoring and explore the potential for the prevalence of DUI on the basis of 2 data sets (DUI and non DUI).

Results: The study results showed that driver error and male drivers were the 2 most significant risk factors for DUI, with the highest scores and potential for prevalence. The nonuse of restraint systems, driver experience, and driver age are the factors with a significant prediction of involvement in an accident and an insignificant prediction of DUI.

Conclusions: Following the development of the logistic prediction models for DUI drivers, testing of the model was conducted for 3 control driver groups: Car, motorcycle, and bicycle. The prediction model with a probability greater than 50% showed that 77% of car drivers were under the influence of alcohol. Similarly, the prediction percentage for motorcyclists and bicyclists amounted to 71 and 67%, respectively. The recommendation of the study is that drivers whose DUI probability is above 50% should be potentially suspected of DUI. The results of this study can help to understand the problem of DUI among specific driver categories and detect DUI drivers, with the aim of creating successful traffic safety policy.     

Analysis of driver injury severity levels at multiple locations using ordered probit models

Problem: The occurrence and outcome of traffic crashes have long been recognized as complex events involving interactions between many factors, including the roadway, driver, traffic characteristics, and the environment. This study is concerned with the outcome of the crash. Method: Driver injury severity levels are analyzed using the ordered probit modeling methodology. Models were developed for roadway sections, signalized intersections, and toll plazas in Central Florida. All models showed the significance of driver's age, gender, seat belt use, point of impact, speed, and vehicle type on the injury severity level. Other variables were found significant only in specific cases. Results: A driver's violation was significant in the case of signalized intersections. Alcohol, lighting conditions, and the existence of a horizontal curve affected the likelihood of injuries in the roadway sections' model. A variable specific to toll plazas, vehicles equipped with Electronic Toll Collection (ETC), had a positive effect on the probability of higher injury severity at toll plazas. Other variables that entered into some of the models were weather condition, area type, and some interaction factors. This study illustrates the similarities and the differences in the factors that affect injury severity between different locations.     

从道路交通事故统计分析对比谈预防措施

笔者通过对比我国与世界上部分发达国家近十年来在道路交通事故方面的统计数据 ,并根据经济发展水平与机动车保有量的正比例关系及当前我国道路交通事故现状 ,分析得出道路交通安全事故的发生 ,与人、车辆、道路、环境信息及管理等因素具有密切关系 ,其中人 (尤其是驾驶员 )作为交通行为的主体 ,是道路交通事故诱因中一个关键性因素。由此 ,作者提出了要有效预防和减少道路交通事故的发生 ,必须将人、车、路、环境信息和管理等诸因素作为一个有机整体系统思考 ,且在未来的道路交通发展中应引入交通稳静化理念 ,以实现道路交通的安全、畅通与高效