A Study Of Association Rule Mining Algorithm Based On MFP-tree

DM technique is a hot topic in database and artificial intelligence researches, bringing itself to the attention of the science community and industrial community. Rule mining Algorithm is one of the important branches in data mining, mainly adopted to discover the correlation connections among data concentrators. Data amount increases dramatically as the fast growth of information technology and most of the databases are relational-like data base. Thus, the study of effective techniques for rule mining algorithm in relational-like database holds vast vistas. FP-Growth, an association rule mining algorithm based on MFP Tree, demanding no Candidate Itemset, is the most used method for mining frequent itemsets. Occupying a great deal of memory, FP-growth algorithm runs slowly and even fails to construct FP-tree based on memory. To solve these problems, this paper proposes a new association rule mining algorithm, DMFP, to meet the demand of LDB mining.The works of this dissertation are as follows:(1) Discusses present researches on data mining and association rule mining techniques.(2) Reviews traditional association rule mining algorithms and their defects after a theoretical study of theories of association rule mining algorithm.(3) Propose an improved association rule mining algorithm fit for LDB mining, DMFP, with experiment test and functional analysis. DMFP reduces the memory usage of FP-tree, and experiments show DMFP demonstrates a better function than FP-Growth.(4) Combining to actual application, DMFP algorithm is used to a personal information on the investigation table. These experimental results show that DMFP algorithm is effective.