18 April 2024 in Transportation Research Record: Journal of the Transportation Research Board
Accurate predictive modeling is often hindered by the prevalent issue of class imbalance within weather-related crash datasets. To address this critical challenge, this study introduces a novel and tailored synthetic data generation technique aimed at effectively handling nominal predictors specific to weather-related crash cases in North Carolina. Data treatment techniques such as the synthetic minority over-sampling technique-nominal (SMOTE-N) and adaptive synthetic-nominal (ADASYN-N) were investigated in this study. A comprehensive comparison of these data treatment techniques is conducted using two prominent machine learning models: the bagging algorithm (random forest [RF]) and the boosting algorithm (extreme gradient boosting [XGBoost]). The findings indicate that the effectiveness of data treatment varies with the severity level and the algorithm used. The ADASYN-N technique was observed to be highly effective for severe and moderate injury crash prediction using both the RF and XGBoost models, while the control dataset and SMOTE-N demonstrated notable performance in property damage only crash prediction using both the RF and XGBoost models. The findings from evaluating the performance of these models with data treatment methods serve as a benchmark for practitioners in selecting appropriate synthetic sample generation techniques, consequently facilitating the development of more accurate crash severity prediction models and contributing to enhanced traffic safety strategies.
10 December 2024 in Sustainable Cities and Society
6 September 2023 in Heliyon
The perception of non-motorists toward autonomous vehicles (AVs) could change or remain the same after hearing about a fatal AV crash. This study aims to discern the differences between non-motorists whose opinions were influenced and those whose views remained unchanged after hearing about the fatal AV crash. Additionally, this study investigates how the operational competence of AV manufacturers affect non-motorist's safety perception. A fixed parameter ordered logit model was explored to explain non-motorists’ safety perception for respondents who changed their opinion after a crash occurrence and those who did not. A random parameter ordered logit model was then explored to address the difference in residual variance. The finding from the random parameter ordered logit model indicates that two factors (i.e., future impact of AVs on traffic injuries and willingness to adopt Pittsburgh's streets as a proving ground) were significant for the respondents with negative change in opinion after the crash. The result also suggests that negative information such as the crash in Arizona may influence non-motorists’ safety perception of AVs, but this effect can be mitigated by providing positive information about the AVs. The results also demonstrate that non-motorists who perceive the operational competence of the AV manufacturers to be high have higher safety perception of the AVs, while those who perceive the operational competence of the AV manufacturers to be low have lower safety perception of the AVs. This study offers recommendations for policies and interventions to enhance non-motorists' safety based on the examination of factors associated with safety perception among respondents who changed their opinion following a crash and those who maintained their initial stance.