Research And Application Of Association Rules Mining Based On FP-tree

Data mining technology is current research in the information science field. Mining association rules from datasets, which is one of the most important research fields in data mining, can reveal the relationships among items. This paper studies the algorithms of association rules mining based on FP-tree.The DMFIA based on the FP-tree algorithm is one of the widely- applied and most efficient algorithms for mining the maximal frequent item-sets currently . But it generates the candidate ones in the maximal frequent item-sets and it will bring on a batch of the candidate sets when the frequent items are very much in database and they are so few in each transaction. To this problem, this paper improved the traditional FP-tree structure and proposed the maximal frequent item-sets mining algorithm based on the improved FP-tree. It needn't to generate the candidate maximal frequent item-sets. The improved FP-tree is unilateralist, and each point save the pointers of the parents', which will economize memory. It is shown in our experimental results that the FP-MFI algorithm is more effectively than the DMFIA also based on FP-tree in the mining maximal frequent item-sets when the frequent items are very much in database and they are so few in each transaction.Most algorithms based on FP-tree possess much memory and run slow in the mining large database. In view of this problem, this paper proposed an algorithm for mining large database based on FP-tree. It utilizes a method to divide database, and it can decompose effectively a FP-tree into many small FP-trees, and is suitable for running in several computers, finally merges each results, and obtains the frequent items completely. It is shown in our experimental results that the FP-LMFI algorithm can distribute the relational rules mining in the large database into each of the hosts without better calculation ability.Finally, this paper discussed the algorithm of FP-MFI for mining the maximal frequent item-sets and the algorithm of FP-LMFI for mining the larger database based on FP-tree applied in the student choosing class system.