Random forest models for motorcycle accident prediction using naturalistic driving based big data.

Motorcycle accident studies usually rely upon data collected from road accidents collected through questionnaire surveys/police reports including characteristics of motorcycle riders and contextual data such as road environment. The present study utilizes big data, in the form of vehicle trajectory patterns collected through GPS, coupled with self-reported road accident information along with motorcycle rider characteristics to predict the likelihood of involvement of a motorcyclist in an accident. Random Forest-based machine learning algorithm is employed by taking inputs based on a variety of features derived from trajectory data. These features are mobility-based features, acceleration event-based features, aggressive overtaking event-based features and motorcyclists socio-economic features. Additionally, the relative importance of features is also determined which shows that aggressive overtaking event-based features have more impact on motorcycle accidents as compared to other categories of features. The developed model is useful in identifying risky motorcyclists and implementing safety measures focused towards them.

The effect of different mobile uses on crash frequency among young drivers: application of statistical models and clustering analysis.

This study focuses on investigating the use of mobile phones among young drivers by employing an online questionnaire survey data. Ordinal logistic regression model was used for modelling the probabilities of crashes due to different uses of mobile phone while driving. Moreover, binary logistic regression models were used for predicting the probabilities of different uses of mobile phone. Logistic regression models revealed that texting and internet use have the same likelihood of causing crashes. Drivers having prior experience of being fined for mobile phone use, also showed a higher tendency to be involved in 2 crashes. Moreover, these drivers had a higher likelihood of being involved in texting, as compared to other uses of mobile phones. Drivers with more education had a higher tendency for internet use during driving. Drivers who use mobile phone for long periods during driving have a lesser tendency to get involved in texting, internet use or GPS navigation. Moreover, drivers with a previous crash record have less likelihood of being involved in texting. The models of this study can be useful in developing effective road safety measures. Clustering was also applied in this study which reinforced the findings of the statistical analysis and models.

Transformation of urban mobility during COVID-19 pandemic - Lessons for transportation planning.

INTRODUCTION: The coronavirus disease (COVID-19) pandemic is a global threat that started in Wuhan, China, in 2019 and spread rapidly to the globe. To reduce the spread of the COVID-19, different non-pharmacological control measures have been conducted in different countries, which include social distancing, distance working, and stay-at-home mandates. These control measures had affected global transportation and mobility significantly. This study investigated the short-term changes in urban mobility, tropospheric air pollution, and fuel consumption in two major cities of Saudi Arabia, namely, Riyadh and Jeddah. METHODS: In this study, the dynamics of the number of trips and trip purposes in different provinces of the country were analyzed, focusing on the pandemic period and the lockdown program. These changes impacted fuel consumption and, consequently, air pollutants. The quantity of fuel consumption and its trend was projected considering a few possible fuel consumption and emission scenarios. It is also expected that fuel price plays a role in fuel consumption. The spatial and temporal distributions of the remote sensed tropospheric Nitrogen Dioxide (NO2) levels in different provinces were presented to depict the short 19 and long-term impact on the air quality due to the changes in mobility. RESULTS: The significant reduction in urban mobility has been observed since the beginning of the first partial curfew in March 2020 compared to that in 2019. The air pollutant levels (such as NO2) in 2020 after the pandemic were generally less than those of 2019. The fuel consumption has been following a decreasing trend in 2020 starting from January due to dynamic fuel price and the additional influence of pandemic. Based on the current online shopping pattern, it is argued that there will be some permanent behavioral changes in urban mobility, which will decrease some shopping trips at least immediately after the recovery from the pandemic. CONCLUSIONS: This study concluded that the availability of global urban mobility data, remote sensed based tropospheric air pollution data, and global fuel consumption database are important sources of information to investigate the impact of COVID pandemic, especially for the developing countries which suffer from scarcity of pertinent urban mobility information. It seems that, at least in the study area, the spread of COVID-19 is a complex phenomenon in which several exogenous factors, in addition to the curfew protocols, affect the spread of the virus.

A Nested Ensemble Approach with ANNs to Investigate the Effect of Socioeconomic Attributes on Active Commuting of University Students.

Analysis of travel mode choice is vital in policymaking and transportation planning to comprehend and forecast travel demands. Universities resemble major trip attraction hubs, with many students and faculty members living on campus or nearby. This study aims to investigate the effects of socioeconomic characteristics on the travel mode choice of university students. A nested ensemble approach with artificial neural networks (ANNs) was used to model the mode choice behavior. It was found that students generally prefer motorized modes (bus and car). A more detailed analysis revealed that teenage students (aged 17-19 years) had an approximately equal probability of selecting motorized and non-motorized modes. Graduate students revealed a higher tendency to select motorized modes compared with other students. The findings of this study demonstrate the need to promote non-motorized modes of transport among students, which is possible by providing favorable infrastructure for these modes.

Modelling of Asphalt's Adhesive Behaviour Using Classification and Regression Tree (CART) Analysis.

The modification by polymers and nanomaterials can significantly improve different properties of asphalt. However, during the service life, the oxidation affects the constituents of modified asphalt and subsequently results in deviation from the desired properties. One of the important properties affected due to oxidation is the adhesive properties of modified asphalt. In this study, the adhesive properties of asphalt modified with the polymers (styrene-butadiene-styrene and styrene-butadiene) and carbon nanotubes were investigated. Asphalt samples were aged in the laboratory by simulating the field conditions, and then adhesive properties were evaluated by different tips of atomic force microscopy (AFM) following the existing functional group in asphalt. Finally, a predictive modelling and machine learning technique called the classification and regression tree (CART) was used to predict the adhesive properties of modified asphalt subjected to oxidation. The parameters that affect the behaviour of asphalt have been used to predict the results using the CART. The results obtained from CART analysis were also compared with those from the regression model. It was observed that the CART analysis shows more explanatory relationships between different variables. The model can predict accurately the adhesive properties of modified asphalts considering the real field oxidation and chemistry of asphalt at a nanoscale.