Association Rule Mining for Accident Record Data in Mines

Abstract — Association Rule Mining is one of the important and well accepted application areas in the field of Data Mining where rules are found between the data items which helps to determine the relationships between the data items. Thus helps to make the prediction for future in a better way. This study examined the group differences in accident susceptibility among underground coal mine workers accounting for different factors associated with them. Data was analysed with a bivariate modelling technique in the initial phase. Knowledge discovery in Database model- A Data mining technique was used for this purpose to take multiple parameters into account simultaneously. Data were collected from eight different underground coal mines for a period of ten years. The case study results revealed that different age of workers bear no significant differences in their accident susceptibility; however, the type of mines and designation show significant differences in their risk of injuries. It is inferred based on the Data Mining technique that among the various reasons of accidents, accident by haulage and conveyor, fall of person and fall of object possesses more risk than the traditional Roof and Side falls. There are other major conclusions which have been deduced from the work Later on various rules were also predicted from the dataset with the required rule accuracy and rule coverage enabling us to visualize and predict the general trend