The Research On The Algorithms Of Mining Association Rules

As one of the important contents in data mining, association rule mining aims to discover the interesting connection or the correlation midst a set of objects in a database. Association rule mining has become a hot research topic in recent years, and it has been used widely in selective marketing, decision analysis and business management.In the thesis, some classical algorithms for mining association rules have been systematically studied and comprehensively summarized. On the basic of previous research, the novel algorithms for mining association rules and incremental updating of association rules are proposed.Firstly, the thesis introduces some basic knowledge of data mining and association rules and some classical algorithms for association rules.Secondly, the thesis analyses the disadvantage of FP-Growth.Taking measures from data structure and mining means, a novel algorithm for mining frequent patterns based on improved compressed FP-tree, i.e. QMFI-ICFP, is proved. This algorithm saves large memory space occupied by FP-tree and the cost of constructing many conditional FP-trees. Experiments show that the time and space for QMFI-ICFP have reduced significantly compared to FP-growth mining.Finally, a new incremental updating algorithm based on matrix for maintaining discovered association rules, i.e. IUBM, used when the transaction database increases and the minimum support changes, is presented in the thesis. Some analysis to the new algorithm is presented. Compared with the other algorithm, the algorithm just scans the new database db once, therefore it has a higher efficiency.