SUMMARY
Occupational injuries and fatalities are one of the most significant issues in the construction industry. Variables, such as workers’ behavior, age, worksite condition, and type of activity, play key roles in the occurrence of construction accidents. In recent years, data mining techniques have been successfully used not only in health, economy, and social sciences but also in construction-related fields. In this study, C5.0 decision tree algorithm was used to analyze the accident data obtained from the Social Security Institution of Turkey. A classification tree model was created to discover the associations between the attributes. The results show the relationship between the injury status of workers and the attributes, and the accuracy rate of the model was 70.26%. Meanwhile, unsafe conditions, type of accident, and activity type were the most important attributes in the model. Furthermore, the predictor importance of the attributes was compared, and several outcomes were discovered; for instance, the workers’ educational background has greater predictive power than age. On the other hand, the branches of the decision tree pointed out several attribute sequences due to their high rated serious/fatal injury rates. The results of this study can be used in the prevention and mitigation strategies for construction accidents.