Abstract: For the inadequacy of Apriori algorithm in association rules,this paper presents two methods of Apriori algorithm based on compression matrix approach.Improved algorithms make full use of the matrix and give compression on it.to significantly reduce the number of scans the database,and improve the generation efficiency of the frequent itemsets.and then improve the efficiency of the algorithm effectively.Meanwhile,the application example and algorithm performance analysis shows that the proposed two improved algorithms are efficient association rule mining method,and the properties are belter than the Apriori algorithm.