Comparative Analysis of the Proportional Distribution Method and the Random Forest Algorithm for Predicting Pedestrian Traffic Accident Risk

The risk of pedestrian-involved traffic accidents represents a significant challenge to road safety and necessitates objective methods for analyzing the contributing factors. This study presents a comparative analysis of two methodologies for predicting the risk of pedestrian traffic accidents: a methodology based on proportional risk distribution and the Random Forest algorithm. The analysis utilizes data derived from real court cases, where linguistic variables defined as risk factors are categorized and quantified based on expert evaluations. The results demonstrate that both approaches are applicable for risk assessment, with Random Forest exhibiting higher accuracy and robustness in handling complex and heterogeneous data. Correlation analysis confirms a statistically significant linear relationship between the outputs of the two methods, supporting their validity. Graphical representations derived from the results offer a visual interpretation of risk severity and facilitate comparison between the two approaches. The proposed method is intended for road safety experts, engineers, analysts, and institutions in the field of transportation safety. Its primary aim is to provide an objective and quantitative tool for evaluating the risk factors contributing to pedestrian-related incidents. The method supports informed decision-making regarding preventive measures and awareness campaigns targeting both drivers and pedestrians.