Assessment of the Spatial and Temporal Dimensions of Traffic Safety: A Statistical and Machine Learning Approach to Modeling Traffic Crash Severity and Collision Types

This dissertation focuses on investigating various critical aspects of traffic safety in Bexar County, Texas, with the aim of developing strategies for preventing crashes and improving overall traffic network efficiency. The first research objective focuses on analyzing traffic crashes, identifying the causes of crashes, and developing targeted interventions for reducing the frequency and severity of traffic crashes. This study utilizes spatial statistical techniques and trend visualization tools to identify high-risk locations and times, including those resulting in fatal or incapacitating injuries. The second research objective investigates the impact of the COVID-19 pandemic on truck crashes. The third research objective involves developing a Machine Learning Multi-Label Classification (MLC) tool for simultaneous classification of collision type and crash injury type. This tool utilizes past crash data for numerical analysis and captures the correlation between injury severity and collision type. The fourth research objective involves conducting county-level data exploration using clustering algorithms from unsupervised machine learning. The findings of this dissertation can assist in developing effective interventions and policies to reduce the frequency and severity of traffic crashes, improve traffic network efficiency, and understand the impact of external factors such as the COVID-19 pandemic. The study highlights the importance of traffic safety research and provides a comprehensive understanding of various aspects of traffic safety.