Abstract: A large number of accidents data have been accumulated as a result of the daily safety checks and hidden troubles investigation. In order to exploit the potential value of the data,achieve the early warning task,vertical data format mining algorithm was applied to mining association rules in the data on high dimensional hidden troubles ,and Kulczynski metric and Imbalance Ratio(IR)were used to reduce the effects of frequency difference of hidden danger on the rules. On this basis, an association rules based model was designed for hidden troubles early warning assessment, and the early warning information was visulized. Finally, a complete enterprise hidden troubles warning process was formed. 53 029 pieces of data on hidden troubles in 130 mechanical manufacturing enterprises in 2013 were taken as an example to verify the feasibility.The results show that the accuracy of the method is 80.62%.

Key words: hidden trouble, potential value, vertical data format, association rule, risk assessment model, visualization