Data Warehouse-based Association Rule Mining Algorithm

Association rule is one of active part of data mining. It has been used to find potential and interested information for customs from databases or data warehouses. On the basis of analyzing the defect of existing association rule algorithms, we propose a series of novel ideas, realize several advanced algorithms and achieve favorable result. The main research works are follows:(1) The concept of the best support and the best confident are put forward in association rule. It can be used to mine the most useful association rules in the certain circumstance and attain favorable result of making decision.(2) An advanced algorithm of mining parallel association rules is proposed. By getting rid of the times of scanning databases or data warehouses and the number of candidate items, we can enhance the efficiency of the algorithm. Moreover, the extensity of new algorithm is better than others.(3) A parallel algorithm for mining weighted association rules is proposed. In order to make the algorithm correspond with the reality, we offer each item a different weight value so that it can represent the importance of individual items from databases or data warehouses. In this way, we may discover the useful association rules for customs.(4) An algorithm of mining quantitative association rules is put forward. Quantitative attribute values are partitioned into basic intervals according to their distribution in the databases or data warehouses, and if possible ,the adjacent basic intervals will be merged. Then the intervals are mapped into Boolean attributes. At last, general interesting quantitative association rules can be mined.Theory analysis and simulation results for data show the result based on these methods are improved much more than normal algorithm's.