摘要
简要描述了加权关联规则问题及离散粒子群优化算法,提出了一种基于粒子群优化(PSO)算法的加权关联规则挖掘算法(PSO-WMAR).实验证明,本算法运行时间更省,产生的规则数更少且更有效.该算法具有以下特点:1)把关联规则挖掘的两个阶段结合在一起,无须先挖掘出全部频繁项目集然后再提取规则;2)只需要扫描一次数据库;3)把兴趣度引入适合度函数之中,挖掘出的规则数量更少、更有效;4)求加权频繁项目集无须查找所有候选加权频繁项目集,或者求频繁项目集的高序子集或非频繁项目集的低序超集.
This paper firstly describes the problem of mining association rules with weighted items and the algorithm of binary particle swarm optimization （ PSO）, and then presents a PSO-based weighted items association rules mining algorithm （PSO-WMAR）. Experiments show that the PSO-WMAR algorithm is effective and can save time in mining of the association rules. It has the following characteristics： 1 ） The algorithm combines the two phases of mining association rules together, and needn＇t mine all the candidate large itemsets before deriving the rules; 2） The algorithm only scan the database once; 3 ） Interesting degree is used to calculate the fitness function of particle swarm optimization, so we can derive less but much more effective rules ; 4） To search a weighted large itemset, we needn＇t find out all their candidate weighted large itemsets. In addition, we needn＇t find out the high-order subset of a weighted large itemset, and check the low-order superset of a weighted small itemset to make certain all large itemsets.
出处
《集美大学学报（自然科学版）》
CAS
2007年第1期52-58,共7页
Journal of Jimei University：Natural Science
基金
福建省自然科学基金资助项目(A0510023)
关键词
粒子群
加权关联规则
数据挖掘
particle swarm
weighted association rules
data mining
