一种基于权重的Apriori改进算法  

An improved Apriori mining algorithm based on weight

作　　者：刘毓[1] 李莎[1] 

机构地区：[1]西安邮电大学通信与信息工程学院,陕西西安710121

出　　处：《西安邮电大学学报》2017年第4期95-100,共6页Journal of Xi’an University of Posts and Telecommunications

基　　金：陕西省工业科技攻关项目(2016GY-113)

摘　　要：针对Apriori算法在频繁项集挖掘过程中的缺陷,提出了一种基于权重的改进Apriori算法。该改进算法通过一次扫描事务数据库构造出二元事务矩阵,再用各事务和各项的平均权重替代权重支持度,最终挖掘出事务库中的频繁项集。通过实例分析和性能测试,证明了改进的Apriori算法避免了重复扫描事务数据库,使得算法在性能上有了明显优化,并且挖掘出了Apriori挖掘不到的、隐藏的、有价值的规则。An improved Apr ior i algorithm based on weight is proposed to deal w ith the defects of Apriori algorithm in frequent itemsets mining. In this improved algorithm , the binary transaction matrix is constructed through a scanning transaction database, and the weight support is then replaced with the average weight of each transaction and each item , and the frequent itemsets is finally digged out in the transaction library . Through the case analysis and performance test, it is proved that the improved Apriori algorithm can avoid the repeated scanning of the transaction database to make the algorithm optimized, and to excavate the hidden, hidden and valuable rules that Apriori cannot find before.

关 键 词：数据挖掘 APRIORI算法 频繁项集 权重支持度 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]