Assessing and predicting crash dynamics with and without road safety measures on the Dejen to Bahir Dar highway in Ethiopia.

Reducing road traffic accidents and enhancing road safety remain pressing concerns for effective transportation systems worldwide. However, in developing countries like Ethiopia, addressing these issues faces significant challenges. Despite the success of cost-effective safety measures in developed countries, similar strategies are often lacking in Ethiopia. This study aims to assess traffic crash patterns, contributing factors, and the effectiveness of safety measures along the Dejen to Bahir Dar highway in Ethiopia. The primary objective is to identify the key determinants of accident frequency and severity and evaluate the impact of safety interventions using advanced statistical models, including Empirical Bayes. The study finds that several factors significantly influence crash occurrence, including vehicle type, crash type, weather conditions, operational factors, road geometry, and driver demographics such as age, sex, and experience. By analyzing these factors, the paper proposes a set of practical engineering solutions, prioritizing high-risk groups like young and inexperienced drivers and advocating for stricter regulations on high-risk vehicles. Further recommendations focus on improving road safety during adverse weather conditions, enhancing road maintenance on straight sections, and implementing better enforcement of speed limits and driver fatigue regulations. Other proposed interventions include the installation of roadside barriers, new traffic signage, and improved pedestrian facilities. This study contributes valuable insights into the determinants of traffic crash in Ethiopia and offers data-driven recommendations for improving road safety. It also outlines future research avenues, such as improving data quality, conducting spatial analyses of crash hotspots, and exploring the influence of environmental and road conditions through advanced statistical, and Empirical Bayes methods.