摘要
传统的Apriori算法由于始终保持单一的最小支持度,所以在实际应用中不能挖掘小比例事件中的关联规则。针对这一缺陷,该文提出并实现了一种基于概率的多最小支持度关联规则算法。该算法针对每个项目设定了最小项支持度,最小项支持度与该项目的出现概率相关。实验证明该算法不仅能有效地挖掘出发生概率较低的事件中的关联规则,同时又不丢失原有的大概率事件中的关联规则。另外,实验结果也说明该算法存在候选项集增多的缺点。
Because the traditional Apriori algorithm always keeps a single mini - support, it can not bring out in the practical application the association rules from the little probability items. To solve the problem, this paper presents and realizes a new algorithm of multi - mini - support association rules based on probability. This new algorithm sets for each item a minimum support rate, which is related to the probability of the item. Experiments show that the new algorithm can not only bring out the association rules from the little probability items, but maintain these rules in the large probability items as well. In addition, the result of the experiments also reveals that the defect of this algorithm is the increase of candidate item sets. KEY-WORDS：
出处
《计算机仿真》
CSCD
2006年第7期115-118,160,共5页
Computer Simulation
基金
云南自然科学基金项目(2002F0013M)
温州职业技术学院重点课题(WZY2005003)
关键词
关联规则
多最小支持度
概率
数据挖掘
算法
Association rule
Multi - mini - support
Probability
Data mining
Algorithm
