快速频繁项集挖掘算法研究-电路与系统专业论文.docxVIP

硕士学位论文摘 硕士学位论文 摘 要 FP．growth算法是当前挖掘频繁项目集算法中应用最广，并且不需要候选集 的一种挖掘关联规则的算法。但是，FP．growth算法在挖掘大型数据库时存在占 用内存大和运行速度慢的缺点。为了克服这些不足，本文基于FP—growth算法提 出了两种新的适合于挖掘大型数据库的关联规则算法：EFP．growth算法和 LFP—grwoth算法。 EFP．growth算法利用项集等价类将关联规则挖掘的项集分成互不相交的子 空间的性质，将一个大型数据库分解成多个投影数据库，依次在每一个投影数据 库上进行约束频繁项集挖掘。算法尤其适合支持度较小时的大型数据库的挖掘。 分析和实验表明EFP．growth算法在挖掘大型数据库时时间和空间的性能上均优 于FP．growth算法。而且，随着数据库规模的增大，EFP．growth算法具有更明显 的优势。 LFP．grwoth算法将原来的搜索空间(格)划分成若干个更小的子空间(子格)， 通过子格间的迭代分解，将对网格P(I)的频繁项集挖掘转化为对多个子网格的并 集进行约束频繁项集的挖掘。实验结果和理论分析表明，LFP．growth算法在挖掘 大型数据库时时间和空间的性能上均优于FP．growth算法。而且，随着数据库规 模的增大或支持度阈值的减少，LFP．growth算法具有更明显的优势。 本文还介绍了EFP．growth算法和LFP．grwoth算法挖掘频繁项集的一个应用 实例。 关键词：数据挖掘；FP．树；频繁项集；等价类；格 II AbstractFP-growth Abstract FP-growth algorithm is one of the currently most popular algorithms for mining association rules without candidate generation．However,it has disadvantages such as lower space utilization rate and slower execution time when mining the large datasets· To overcome these drawbacks，based on the FP—growth algorithm，this paper proposed two neW algorithms for mining association rules from large datasets--EFP’growth algorithm and LFP-growth algorithm． Aceording to characteristic of large databases，inspired by the fact that the FP-growth provides an effective algorithm，a new EFP—growth for mining frequent patterns in large databases is proposed．Based on the characteristic of equivalent classes．which separate item sets of association rules into many subsets，proposed algorithm divides a large database into many projection subsets and carries out constrained frequent．Experiments show that the algorithm has accelerated the mining speed and the performance of space scalability is superior to the FP—growth algorithm· Moreover,the algorithm has a very good time and space scalability with the increasing size of database． The neW LFP．growth algorithm divides a large lattice into many sub-lattices，and carries out constrained fequent．Through iterativing decomposition of sublattices， frequent itemset mining in lat