This thesis explores factors that influence traffic accidents at both signaled and non-signalized intersections in North Carolina. This follows from a recent Masters’ thesis [4]on bicyclist injury severity at unsignalized intersections in North Carolina. We revisitthe original data for our analysis, increase the number of observations and find that onlythree injury’s levels can be correctly classified (83% accuracy rate) using a neural networkmachine learning algorithm. With this information, we examine the relation between Ambulance and Injury levels and study the effect of traffic control (or its absence) on injury levels. We conclude that there is an increase in odds in the severity of the injury level in the absence of a traffic control. Further, given the (3) levels of injuries, we find that Road Speed Limit, Driver’s Estimated Speed Limit, Light Conditions (daylight or not) and Road Characteristics (curve or straight) are statistically significant factors in a multinomial logistic regression model.
Date
2020-01-01
Department/Academic Units
Department of Mathematics, Computer Science, and Engineering Technology
