MS-Miner: A New Frequent Itemsets Mining Algorithm

Frequent itemsets mining can be widely used in mining association rules, correlation analysis, intrusion detection, sequence mode analysis, classification, clustering and other data mining tasks. So far there have been many efficient frequent itemsets mining algorithms. In this thesis, many frequent itemsets mining algorithms are studied in depth. Especially, transaction data base's storage structure and some effective techniques in different growth algorithms are analyzed systematically. And then, a new algorithm is introduced.Firstly, we analyze the concepts related association rules and frequent itemsets mining, the present research and the facing problems, study several typical frequent itemsets mining algorithms and compare their respective advantages and application conditions.Secondly, the data storage structures and the effective techniques of a large number of algorithms have been studied detailedly, and the data structures and strategies of the various growth algorithms are analyzed specially.Finally, the three applications--FP-Tree, FP-Array and Bitmap-Count are explained in detail. Based on the three techniques, a new frequent itemsets mining algorithm - MS -Miner is designed.Experimental results show that our method is an efficient algorithm, not only in the working efficiency but also in the memory consumption and scalability.