Emmanuel Kidando

Evaluating Aging Pedestrian Crash Severity with Bayesian Complementary Log–Log Model for Improved Prediction Accuracy

Reliable prediction accuracy is an essential attribute for crash prediction models. Generally, more severe injury outcomes, such as fatalities, are rarer than less severe crashes, such as property damage only or minor injury crashes. The complementary log–log (cloglog) model, commonly used in epidemiological research, is known for its accuracy in predicting rare events. This study implemented the cloglog model in analyzing pedestrian injury severity and compared its performance with the two conventional models used in injury severity research: the probit and logit models. The three models were developed with data from 1,397 crashes involving aging pedestrians that occurred in Florida from 2009 through 2013. The response variable, injury severity level, was binary and categorized as either fatal or severe injury or minor or no injury. The study used three accuracy metrics (deviance information criteria, prediction accuracy, and receiver operating characteristics curves) to compare the performance of the models. The cloglog model outperformed the probit and logit models in overall goodness of fit and prediction accuracy. More important, the cloglog model outperformed the other two models considerably for predicting fatal and severe crashes according to the recall metric (72% accuracy versus 43% and 41% for probit and logit models, respectively). However, the other two models outperformed the cloglog model in predicting crashes with no or minor injuries. Of predictor variables included in the model, six were found to significantly influence fatal or severe injuries for aging pedestrians at 95% Bayesian credible interval. These variables included pedestrian age, alcohol involvement, first harmful event, vehicle movement, shoulder type, and posted speed.

Bayesian Nonparametric Model for Estimating Multistate Travel Time Distribution

Multistate models, that is, models with more than two distributions, are preferred over single-state probability models in modeling the distribution of travel time. Literature review indicated that the finite multistate modeling of travel time using lognormal distribution is superior to other probability functions. In this study, we extend the finite multistate lognormal model of estimating the travel time distribution to unbounded lognormal distribution. In particular, a nonparametric Dirichlet Process Mixture Model (DPMM) with stick-breaking process representation was used. The strength of the DPMM is that it can choose the number of components dynamically as part of the algorithm during parameter estimation. To reduce computational complexity, the modeling process was limited to a maximum of six components. Then, the Markov Chain Monte Carlo (MCMC) sampling technique was employed to estimate the parameters’ posterior distribution. Speed data from nine links of a freeway corridor, aggregated on a 5-minute basis, were used to calculate the corridor travel time. The results demonstrated that this model offers significant flexibility in modeling to account for complex mixture distributions of the travel time without specifying the number of components. The DPMM modeling further revealed that freeway travel time is characterized by multistate or single-state models depending on the inclusion of onset and offset of congestion periods.

Evaluating Recurring Traffic Congestion using Change Point Regression and Random Variation Markov Structured Model

This study develops a probabilistic framework that evaluates the dynamic evolution of recurring traffic congestion (RTC) using the random variation Markov structured regression (MSR). This approach integrates the Markov chains assumption and probit regression. The analysis was performed using traffic data from a section of Interstate 295 located in Jacksonville, Florida. These data were aggregated on a 5-minute basis for 1 year (2015). Estimating discrete traffic states to apply the MSR model, this study established a definition of traffic congestion using Bayesian change point regression (BCR), in which the speed–occupancy relationship was explored. The MSR model with flow rate as a covariate was then used to estimate the probability of RTC occurrence. Findings from the BCR model suggest that the morning peak congested state occurs once speed is below 58 miles per hour (mph), whereas the evening peak period occurs at a speed below 55 mph. Evaluating the dynamics of traffic states over time, the Bayesian information criterion confirmed the hypothesis that a first-order Markov chain assumption is sufficient to characterize RTC. Moreover, the flow rate in the MSR model was found to be statistically significant in influencing the transition probability between the traffic regimes at 95% posterior credible interval. The knowledge of RTC transition explained by the approaches presented here will facilitate developing effective intervention strategies for mitigating RTC.

Evaluating Factors Influencing the Severity of Three-Plus Multiple-Vehicle Crashes using Real-Time Traffic Data

Multiple-vehicle crashes involving at least two vehicles constitute over 70% of fatal and injury crashes in the U.S. Moreover, multiple-vehicle crashes involving three or more vehicles (3+) are usually more severe compared with the crashes involving only two vehicles. This study focuses on developing 3+ multiple-vehicle crash severity models for a freeway section using real-time traffic data and crash data for the years 2014–2016. The study corridor is a 111-mile section on I-4 in Orlando, Florida. Crash injury severity was classified as a binary outcome (fatal/severe injury and minor/no injury crashes). For the purpose of identifying the reliable relationship between the 3+ severe multiple-vehicle crashes and the identified explanatory variables, a binary probit model with Dirichlet random effect parameter was used. More specifically, Dirichlet random effect model was introduced to account for unobserved heterogeneity in the crash data. The probit model was implemented using a Bayesian framework and the ratios of the Monte Carlo errors were monitored to achieve parameter estimation convergence. The following variables were found significant at the 95% Bayesian credible interval: logarithm of average vehicle speed, logarithm of average equivalent 10-minute hourly volume, alcohol involvement, lighting condition, and number of vehicles involved (3, or >3) in multiple-vehicle crashes. Further analysis involved analyzing the posterior probability distributions of these significant variables. The study findings can be used to associate certain traffic conditions with severe injury crashes involving 3+ multiple vehicles, and can help develop effective crash injury reduction strategies based on real-time traffic data.

Likelihood estimation of secondary crashes using Bayesian complementary log-log model

Secondary crashes (SCs) occur within the spatial and temporal impact range of a primary incident. They are non-recurring events and are major contributors to increased traffic delay, and reduced safety, particularly in urban areas. However, the limited knowledge on the nature of SCs has largely impeded their mitigation strategies. The primary objective of this study was to develop a reliable SC risk prediction model using real-time traffic flow conditions. The study data were collected on a 35-mile I-95 freeway section for three years in Jacksonville, Florida. SCs were identified based on travel speed data archived by the Bluetooth detectors. Bayesian random effect complementary log-log model was used to link the probability of SCs with real-time traffic flow characteristics, primary incident characteristics, environmental conditions, and geometric characteristics. Random forests technique was used to select the important variables. The results indicated that the following variables significantly affect the likelihood of SCs: average occupancy, incident severity, percent of lanes closed, incident type, incident clearance duration, incident impact duration, and incident occurrence time. The study results have the potential to proactively prevent SCs.