Analysis of distracted driving crashes in New Jersey using mixed logit model

Interpretable machine learning for evaluating risk factors of freeway crash severity

Machine learning (ML) models are widely employed for crash severity modelling, yet their interpretability remains underexplored. Interpretation is crucial for comprehending ML results and aiding informed decision-making. This study aims to implement an interpretable ML to visualize the impacts of factors on crash severity using 5 years of freeways data from Iran. Methods including classification and regression trees (CART), K-nearest neighbours (KNNs), random forest (RF), artificial neural network (ANN) and support vector machines (SVM) were applied, with RF demonstrating superior accuracy, recall, F1-score and ROC. The accumulated local effects (ALE) were utilized for interpretation. Findings suggest that light traffic conditions (volume / capacity < 0.5 ) with critical values around 0.05 or 0.38, and higher proportion of large trucks and buses, particularly at 10% and 4%, are associated with severe crashes. Additionally, speeds exceeding 90 km/h, drivers younger than 30 years, rollover crashes, collisions with fixed objects and barriers, nighttime driving and driver fatigue elevate the likelihood of severe crashes.

Modeling spatiotemporal interactions in single-vehicle crash severity by road types

INTRODUCTION

Spatiotemporal correlations have been widely recognized in single-vehicle (SV) crash severity analysis. However, the interactions between them are rarely explored. The current research proposed a spatiotemporal interaction logit (STI-logit) model to regression SV crash severity using observations in Shandong, China.

METHOD

Two representative regression patterns-mixture component and Gaussian conditional autoregression (CAR)-were employed separately to characterize the spatiotemporal interactions. Two existing statistical techniques-spatiotemporal logit and random parameters logit-were also calibrated and compared with the proposed approach with the aim of highlighting the best one. In addition, three road types-arterial road, secondary road, and branch road-were modeled separately to clarify the variable influence of contributors on crash severity.

RESULTS

The calibration results indicate that the STI-logit model outperforms other crash models, highlighting that comprehensively accommodating spatiotemporal correlations and their interactions is a recommended crash modeling approach. Additionally, the STI-logit using mixture component fits crash observations better than that using Gaussian CAR and this finding remains stable across road types, suggesting that simultaneously accommodating stable and unstable spatiotemporal risk patterns can further strengthen model fit. According to the significance of risk factors, there is a significant positive correlation between distracted diving, drunk driving, motorcycle, dark (without street lighting), and collision with fixed object and serious SV crashes. Truck and collision with pedestrian significantly mitigate the likelihood of serious SV crashes. Interestingly, the coefficient of roadside hard barrier is significant and positive in branch road model, but it is not significant in arterial road model and secondary road model.

PRACTICAL APPLICATIONS

These findings provide a superior modeling framework and various significant contributors, which are beneficial for mitigating the risk of serious crashes.

A temporal analysis of crash injury severities in multivehicle crashes involving distracted and non-distracted driving on tollways

Distracted driving is a prominent cause of traffic crashes and may increase the severity of collisions. Due to the larger speeds on toll ways, distracted driving crashes are more severe than on other types of roads, making it worthwhile to investigate. This study examined the variation in the influence of factors affecting injury severity in crashes involving distracted and non-distracted driving, as well as the change over time, using crash data from Florida toll ways from the 2017 to 2019. A series of random parameters logit models with heterogeneity in the means and variances were developed to analyze different driver-injury severities (no injury, minor injury, and severe injury) in crashes involving distracted and non-distracted driving. In addition, likelihood ratio tests were conducted to determine whether model parameters differed between different driver behaviors (distracted/non-distracted driving) and among years. Several factors potentially impacting injury severities were studied, including driver, crash, vehicle, roadway, environment, temporal, and others. Significant disparities were observed between the contributing factors of the severity of crashes involving distracted and non-distracted driving. Results showed that considerable differences were also observed in the severity of injuries caused by two types of crashes and distracted driving resulted in more serious crashes than non-distracted driving. Despite model results indicated that factors influencing injury severity altered over time, several factors, such as motorcycle involvement and commercial car involvement, still exhibited relative temporal stability in non-distracted driving crashes over the three years. Temporal instability and non-transferability were also captured by out-of-sample predictions to verify the temporal shifts of contributing variables from year to year. This study is useful for distinguishing the influence mechanisms between the two types of crashes involving distracted and non-distracted driving, and the results can be applied for safety countermeasures development.

Identifying distracted-driving events from on-road observations using a moving vehicle: A case study in New Jersey

Distracted driving is a major traffic safety concern in the USA. To observe and detect distracted-driving events, various methods (e.g., surveys, videos, and simulations) involving the collection of cross-sectional data from individual subjects have been used in the transportation field. In this study, we employed an unconventional approach of on-road observations using a moving vehicle to collect data on distracted-driving events for multiple subjects in New Jersey. A data-collection crew member continuously navigated selected corridors to record driver-distraction events. A GPS (Global Positioning System) tracker was used to timestamp and record the location of each incident. Two non-parametric tests (Mann-Whitney U test and Kruskal-Wallis test) were performed to identify the significance of the variations in distracted-driving behaviors due to changes in temporal variables (e.g., day of the week, season), the type of roadway, and the geometric properties of the roadway. The results indicated that cellphone use was the leading type of distraction. Additionally, "handheld phone use (phone to ear)," "fidgeting/grooming," "drinking/eating/smoking," and "talking to passengers" events were significantly affected by the time of day and the geometric properties of the roadway. The results of this study are expected to assist state and local agencies in promoting awareness of distracted driving with the aim of reducing the frequency and severity of distracted driving-related crashes.