Toward Safer Work Zones: Predictive Model of Construction Risk of Work Zone on Taiwan’s Freeways

This study analyzes the risk factors associated with engineering vehicle accidents in freeway construction zones in central Taiwan, based on 2,952 accident and environmental records collected from 2021 to 2024. Using Zero-Inflated models, Random Forest (RF), Extreme Gradient Boosting (XGBoost), and SHAP (SHapley Additive exPlanations) methods, key variables influencing accident occurrence were identified and interpreted. A decision tree model was then employed to establish high-risk conditions. The results indicate that cross slope, traffic volume, number of lanes, and radius of curvature are the primary influencing factors.