Research Of Data Mining Algorithm Based On Association Rules

With the development of modern Intelligent system and the accumulation of massive data,people begin to pay more attention to data mining.An important research focus in data mining is the mining of association rules,its purpose is to found all the frequent itemsets in the database,so as to explore the possible potential relationship between each item in the database.Because of the continuous development of database,the existing data mining algorithms of association rules can not adapt to these changes,so it is necessary to study the higher efficient mining algorithm.Through the analysis about the data mining algorithms of association rules,the thesis proposes three different improved algorithms based on the existing research.The main contents of this thesis are as follows:(1)Introduces the association rules about the basic concept,nature,classification and common mining algorithm briefly,deeply analyzes the commonly used data mining algorithms of association rules,including the Apriori algorithm and FP-growth algorithm,and points out the existing shortcomings of the algorithms.(2)Propose the improved algorithm PPV based on the node-list aiming at the shortcomings of the FP-growth algorithm,PPV algorithm constructs the frequent pattern tree structure called PPC-tree and traverse the PPC-tree structure to find the node-list of each node,using the node-list for data mining of association rules.The experimental results show that the improved algorithm has better performance for mining association rules of sparse data.(3)To solve the problem that the efficiency of PPV algorithm mines intensive data is not high,propose the improved algorithm PPV based on the nodesets after deeply analyzes the FP-growth algorithm,constructs the frequent pattern tree structure called POC-tree and traverse the POC-tree structure to find the nodesets of each node,using the nodesets for data mining of association rules.The experimental results show that the improved algorithm has better performance for mining association rules of intensive data.(4)To solve the problem that update the frequent itemsets of association rules,after deeply studyed the incremental update algorithm FUP,proposed an improved incremental updating algorithm IFUP.The algorithm makes full use of the existing mining results,effectively solve the problem that updating the frequent itemsets of association rules.The Experimental results show that the improved algorithm IFUP better than FUP algorithm in performance.