Willingness discrimination method for road traffic accidents based on data reorganization coding and machine learning

Road traffic accidents pose the greatest threat to public safety, and research on pre-incident risk identification based on post-event feature recognition has become a hot topic. This study proposed a method discerning the causes of road traffic accidents based on data reorganization and encoding combined with machine learning techniques. Four distinctive machine-learning models were used to demonstrate the impact of different modeling approaches on the performance of traffic accident analysis models. Leveraging procedural data from a specific city spanning 2015 to 2019, it was observed that after reorganization encoding, the accuracy of machine learning improved by approximately 0.230. The Random Forest model exhibited the highest accuracy, reaching 83.45%. Reorganization encoding proved to be effective in capturing relationships and patterns within the data while reducing feature redundancy. The classification and prediction of road traffic accidents provide a scientific basis for traffic safety management.