Research On Algorithm Of Mining Association Rules

As an important research topic in data mining, association rules mining is widely applied in various fields. Association rules may both examine the knowledge pattern formed for a long time in the profession and discover the secret new rules. Association rules mining has become a top topics for its concise style and easy understanding. This thesis studied and analyzed the data mining technique systematically and deeply, especially for association rules. The main works as the following:Firstly, data mining technology are discussed, including some basic concepts of data mining technology, the current applications and directions of development of data mining technique. Then algorithms of association rules mining is studied deeply. The basic concepts and property are introduced roundly and its typical mining algorithms and these algorithms'basic ideas are summarized and analyzed. The classical algorithms—the Apriori algorithm and some methods of improving Apriori algorithms are analyzed and studied.After analyzing the issue of mining association rules, we design two new algorithms of mining association rules- one is an improved association rule mining algorithm- CRApriori (Combination Reduce Apriori), the other is algorithm of mining association rules based on Matrix. Algorithm of CRApriori optimized the process of pruning of Apriori algorithm and the time of scan transaction database is decreased. An algorithm based on Matrix needs only one pass over database, accelerate the verification speed of the frequent K- item sets, reduce I/O overheads greatly and reduce the store space. Durining making association rules, checking association rule of min_frequent itemset, we can get other frequent itemset association rule including this itemset, so it can reduce the calculating greatly. We compare the two new methods with Apriori algorithm by experiment, and find that they are effective methods of association rulers mining. Finally, we apply the algorithm of mining association rules to university teaching. From teaching value data, we find relation problem of classroom teaching effect and teachers'states.