Prediction of Drunk Drivers Patterns Using the Machine Learning CHAID Algorithm.

Factors influencing drunk driving were identified to help prevent fatal accidents using the CHAID decision tree algorithm. Data from the 2019-2021 KCHS were analyzed, focusing on gender, economic status, education, health, and social factors. The CHAID decision tree identified seven significant risk factors: sex, heavy drinking, walking habits, smoking, subjective health, economic status, and education. For adults, significant predictors included being male, frequent heavy drinking, and lower education. Node 25 showed the highest drunk driving incidence at 16.7%, node 12 the lowest at 1.4%. For the elderly, heavy drinking, lack of physical activity, and smoking were significant, with node 35 at 28.9% incidence and node 22 at 0.0%. For adults, being male, heavy drinking, and lower education were key risks, while for the elderly, heavy drinking, lack of physical activity, and smoking were the highest risks.