Exploring spatial heterogeneity in factors associated with injury severity in speeding-related crashes: An integrated machine learning and spatial modeling approach.

Speeding, a risky act of driving a vehicle at a speed exceeding the posted limit, has consistently emerged as a leading contributor to traffic fatalities. Identifying the risk factors associated with injury severity in speeding-related crashes is essential for implementing countermeasures aimed at preventing severe injury incidents and achieving Vision Zero goals. With the wealth of traffic crash data collected by various agencies, researchers have a valuable opportunity to conduct data-driven studies and employ various modeling methods to gain insights into the correlated factors affecting injury severity in traffic crashes. Machine learning models, owing to their superior predictive power compared to statistical models, are increasingly being adopted by researchers. These models, in conjunction with interpretation techniques, can reveal potential relationships between crash injury severity and contributing factors. Traffic crashes are inherently tied to geographic locations, distributed across road networks influenced by diverse socioeconomic and geographical factors. Recognizing spatial heterogeneity in traffic safety is crucial for tailored safety measures to address speeding-related crashes, as a one-size-fits-all approach may not work effectively everywhere. However, most existing machine learning models are unable to incorporate the spatial dependency among observations, such as traffic crashes, which hinders their ability to uncover spatial heterogeneity in traffic safety. To address this gap, this study introduces the Geographically Weighted Neural Network (GWNN) model, a spatial machine-learning model that integrates neural network (NN) and geographically weighted modeling approaches to investigate spatial heterogeneity in speeding-related crashes. Unlike the traditional NN model, which trains a single set of model parameters for all observations, the GWNN trains a local NN model for each crash location using a spatially weighted subsample of nearby crashes, allowing for the quantification of corresponding local effects of features through calculating local marginal effects. To understand the spatial heterogeneity in speeding-related crashes, this study extracted two years (2020 and 2021) of speeding-related crash data from Alabama for the development of the GWNN local models. The modeling results show significant spatial variability among several factors contributing to injury severity in speeding-related crashes. These factors include driver condition, vehicle type, crash type, speed limit, weather, crash time and location, roadway alignment, and traffic volume. Based on the GWNN modeling results, this study identified three types of spatial variations in relationships between contributing factors and crash injury severity: consistent positive associations, consistent negative associations, and inverse associations (i.e., marginal effects can vary between positive and negative depending on the location). This study contributes by integrating advanced machine learning and spatial modeling approaches to uncover intricate spatial patterns and factors influencing injury severity in speeding-related crashes, thereby facilitating the development of targeted policy implementations and safety interventions.

Post-pandemic shared mobility and active travel in Alabama: A machine learning analysis of COVID-19 survey data.

The COVID-19 pandemic has had unprecedented impacts on the way we get around, which has increased the need for physical and social distancing while traveling. Shared mobility, as an emerging travel mode that allows travelers to share vehicles or rides has been confronted with social distancing measures during the pandemic. On the contrary, the interest in active travel (e.g., walking and cycling) has been renewed in the context of pandemic-driven social distancing. Although extensive efforts have been made to show the changes in travel behavior during the pandemic, people's post-pandemic attitudes toward shared mobility and active travel are under-explored. This study examined Alabamians' post-pandemic travel preferences regarding shared mobility and active travel. An online survey was conducted among residents in the State of Alabama to collect Alabamians' perspectives on post-pandemic travel behavior changes, e.g., whether they will avoid ride-hailing services and walk or cycle more after the pandemic. Machine learning algorithms were used to model the survey data (N = 481) to identify the contributing factors of post-pandemic travel preferences. To reduce the bias of any single model, this study explored multiple machine learning methods, including Random Forest, Adaptive Boosting, Support Vector Machine, K-Nearest Neighbors, and Artificial Neural Network. Marginal effects of variables from multiple models were combined to show the quantified relationships between contributing factors and future travel intentions due to the pandemic. Modeling results showed that the interest in shared mobility would decrease among people whose one-way commuting time by driving is 30-45 min. The interest in shared mobility would increase for households with an annual income of $100,000 or more and people who reduced their commuting trips by over 50% during the pandemic. In terms of active travel, people who want to work from home more seemed to be interested in increasing active travel. This study provides an understanding of future travel preferences among Alabamians due to COVID-19. The information can be incorporated into local transportation plans that consider the impacts of the pandemic on future travel intentions.

Are first responders prepared for electric vehicle fires? A national survey.

Transitioning to electric vehicles (EVs) will create both opportunities and challenges. Although some programs and resources related to EVs have been made available to first responders, it remains unknown whether our first responders are well prepared for traffic incidents that involve EVs and whether there are any organizational and geographic disparities in preparedness. To answer these questions, a national survey was conducted to collect feedback on responders' incident management knowledge and training experiences related to EVs, as well as their attitudes and concerns towards EVs. Over 1000 first responders across the country participated in the survey, and the survey had representation from all 50 states and Washington DC. Over 40% of first responders reported never receiving EV-related safety training. Paramedics or EMS are associated with the highest odds of not receiving EV-related training, followed by law enforcement. Geographically, FEMA Region 8 (e.g., Montana and North Dakota) is associated with the highest percentage of not receiving EV training. Regarding EV fire tactics, more than half (57%) of law enforcement officers said they do not know any; responders from towing & recovery also have little knowledge compared to firefighters. Statistical modeling was conducted to explore correlates of responders' EV safety training and knowledge of EV fire tactics. The survey also provided insights about the challenges and risks of managing EV-involved incidents. In summary, responders are greatly concerned about the risks that EVs can pose to their community, and actions must be taken now.