Logistic Model for Predicting Traffic Accidents: An Analysis Based on Labor and Behavioral Factors

Abstract

This study analyzes factors influencing the likelihood of traffic accidents using a logistic regression model. The dependent variable, "traffic accidents in the last year" (yes/no), was modeled based on daily working hours, driving hours, traffic fines in the last year, and active breaks. Results show that daily driving hours and the absence of fines significantly increase the likelihood of accidents, while daily working hours and active breaks have a protective effect. The model demonstrated reasonable fit (Pseudo R2R^2R2 = 0.234) and adequate predictive capacity (AUC = 0.808). These findings underscore the importance of active breaks and effective work time management in mitigating traffic risks.