Using data mining techniques to analyze the influencing factor of unsafe behaviors in Chinese underground coal mines

Abstract

Most coal mine accidents are caused by unsafe behaviors in the production process. Furthermore, there are large quantities of unsafe behavioral data generated in coal mines each year, and the relationships hidden in these data are complex; as a result, finding useful information in these data requires deep analysis. Therefore, to effectively predict and reduce the occurrence of unsafe behaviors in Chinese underground coal mines, this paper analyzes 35,364 unsafe behavioral data from 2220 people during the period from 2013 to 2015 using two methods (association-rule and decision tree) of data mining. The results indicate that the training, attendance, experience and age are the main four factors that affect the frequency of unsafe behaviors, in which the training factor has the greatest impact on unsafe behavior. Moreover, people who have unqualified training, ineffective attendance and less work experience are more likely to conduct unsafe behavior. In addition, six strong association rules about the moderate level of unsafe behavior are excavated using the Apriori algorithm in data mining, and the efficiency of unsafe behavior inspection can be improved by using these strong associations.