Research Of Association Rule Mining Algorithm Based On FP Tree

The quick development of information technology leads an incredible rocket in all kinds of data. Data Mining provides people a new intelligent approach to understand data. FP-Growth algorithm base on FP-tree is one of the data mining algorithms from frequent itemsets , which also is one of the current most popular mining association rules algorithms without candidate itemsets. However, it has disadvantages such as higher memory space occupation , slower execution time and sometimes FP-tree can not be made when mining large databases. To overcome these drawbacks, the dissertation proposes two new mining association rules algorithms for large databases—DFP-Growth1 and DFP-Growth2. The main works in this dissertation are as following:1. The research contents of data mining are summarized. We make an inquiry into the present situation of association rules data mining.2. The theory of association rules data mining is studied and carried out. Some problems about traditional association rules data mining algorithms are discussed.3. Two improved association rules data mining algorithms (DFP-Growth1 and DFP-Growth2) for large databases are proposed based on deep study of FP-Growth algorithms. Experiments have been conducted to test and verify algorithms, by which functions of algorithms also are analyzed.4. New algorithms are combined to actual application. These experimental results show that new algorithms are effective and feasible.