Road traffic accidents have been concerned by the traffic managements and researchers for a long time.Recent years the traffic safety situation in china has been controlled,which stops further deterioration.However,the situation is not optimistic and the number of accidents is still staying at a high level.The existing researches on traffic accidents mainly focused on the accident on urban road and high-grade highway.The traffic security problems on low-grade highway has been ignored.To improve the safety on low-grade highways,a statiscal technique is used to discovery the accident distribution characteristics on it and a multimensional association rules is applied to mine the relevance in the accident database.Based on the characteristics and rules,the low-grade highway accident cause can be found and the targeted measres can be proposed and taken.To apply association rule analysis to traffic accident research,several works are done to enhace the suitability of the method,such as building a multidimensional data model,optimizating the classical Apriori algorithm and strengthening subjective constraints.A six-step data mining flow is put forward,which includes data pre-processing,model building,data mining,reading and applying the founded rules and so on.Meanwhile,the emphases and methods of every step are introduced clearly.Furthermore,a case study based on the accident data on low-grade highway in Suichang Country,Lishui City,Zhejiang Province is shown.And the analysis environment is SQL Server Aanlysis Server,in which a multimensional model with three levels and 26 related attributes is shown.Ultimately,seven attribute combinations which easily causes accident in low-grade highway are found. |