Analysis of risk factors for DUI and DWI crashes considering the built environment.

The risk level of alcohol-involved traffic crashes is closely related to alcohol consumption. However, research on different influencing factors for DUI (Driving Under Influence) and DWI (Driving While Intoxicated) remains limited. This study analyzed data from 3,365 alcohol-related traffic crashes in Tianjin, China. The crashes were categorized into DUI and DWI based on drivers' Blood Alcohol Concentration. Four machine learning models were enhanced and compared. The accuracy, precision, recall and F1-score were used to evaluate the performance of the models. Shapley additive explanations were used to interpret model outputs and quantify risk factors and interaction effects on DUI and DWI crashes. The enhanced CatBoost model performed the best, with an AUC-ROC value of 0.953. The time period of crashes, intersection control or not, and the density of companies were identified as significant factors affecting DUI and DWI crashes. Interaction analysis indicated that drivers aged between 40 and 50 had a higher risk of DWI in areas with high intersection density; two-wheeled motorcycle riders exhibited higher DWI risk compared to car drivers between 21:00 and 24:00. These findings provide valuable insights for the traffic management department to implement targeted and refined control measures for DUI and DWI violations.