Study On Fast Algorithm For Generating Association Rules

Data mining currently is the research frontier within the information science field. It had success applications in many areas. It can find the potential knowledge which hides behind the large data to forecast the trend of things development. The knowledge has greatly improved the ability for decision supporting. Association rule, which has broad applications and caught people's attention, is one of important research areas in data mining. The goal of association rule mining is to discover previously unknown, interesting relationships among attributes from large databases. At present, the research of association rule mining algorithm focuses on improving the efficiency of finding frequent itemset, but has less research on getting association rules from frequent itemset. In this paper we put the emphasis on the latter. The main work is as follows:1. The classical algorithm and Xiongfei Li algorithm, which generate association rules by frequent itemset, are shown. The Boundary algorithm which get the maximal elements of a lower segment is used to get all consequents of the association rules. The classical algorithm and Xiongfei Li algorithm adopt breadth-first strategy to find all consequents layer by layer, Boundary algorithm is depth-first. These three algorithms are compared. The excellence and disadvantage of them are analyzed.2. The GRSET(Generate Rules by using Set-Enumeration Tree) algorithm is proposed based on the research of predecessor. The GRSET algorithm uses the structure of Set-Enumeration Tree and depth-first method to find all consequents of the association rules one by one and gets all association rules correspond to the consequents. The process of GRSET algorithm running was shown and the excellence of it was analyzed.3. Xiongfei Li algorithm, Boundary algorithm and GRSET algorithm were compared by experiments. Experiments show that GRSET algorithm is more efficient than the other two algorithms. At last, analyzing the experiments result, giving the reason why GRSET algorithm is more efficient.