Minimum Support Association Rule Mining

Comparing with the traditional statistic and query methods, Data Mining is a cross-subject which is formed by artificial intelligence, pattern recognition, database, machine learning and management information system, etc. Data mining is a newly-established frontier subject. It has been used extensively, and its application future is also bright.Association rules mining is summarized in this paper. Multiple minimum supports association rules mining is especially researched and analyzed. The main contents are listed as follows:Research and analysis of association rules is presented. Firstly, the basic concepts of association rules mining is summarized. Then association rules mining algorithms are discussed respectively; certain typical algorithms among them are also analyzed. Secondly, the basic theories, mining algorithms and actually research of multiple minimum supports association rules mining are researched and analyzed.For multiple minimum supports association rules mining in the dissertation. We propose a new tree structure multiple item support tree (MIS-tree), to store the crucial information about frequent patterns. Meanwhile, an efficient MIS-tree-based algorithm for mining multiple minimum supports association rules, called the CFP-growth algorithm, is developed for mining all frequent itemsets.For the difficult question to set multiple minimum supports in the dissertation, we propose an efficient algorithm which can maintain the MIS-tree structure. It need not to rescan database and only need to run the mining algorithm repeatedly to adjust item's supports until an appropriate thresholds for all items at a time.Based on the synthetic data, the performance of CFP-growth algorithm is experimented, which is compared with Apriori algorithm, MSapriori algorithm and FP-growth algorithm. The results of the experiments are analyzed, which shows that CFP-growth algorithm is more efficient than the typical MSapriori algorithm. Furthermore the maintaining algorithm of the MIS-tree structure is experimented. The result shows that MIS-tree maintenance method is able to save more time than reconstructing the MIS-tree.