Analyzing pilot-related accidents and incidents by data mining

Identifying patterns of factors resulted in aircraft accidents is important to improve the level of the aviation safety. In this paper we applied contrast set mining technique to study 14 years pilot related accidents and incidents, and identified patterns of factors that are significantly associated with accidents than incidents, and patterns of factors that are more likely lead to accident actually than expected by using attribute focusing technique. Finally identified more interesting patterns of factors that were consistent across the analyses by the two data mining techniques.