一种改进的Apriori关联规则挖掘算法_英文_.doc

一种改进的 Apriori 关联规则挖掘算法 张广路1 ,雷景生2 ,吴兴惠1 (1 . 海南师范大学 数学与统计学院 ,海南 海口 571158 ; 2 . 南京邮电大学 信息与技术学院 ,江苏 南京 211815) 摘 要 :关联规则挖掘是数据挖掘中的一个重要研究内容。为了高效、快速地从事务数据库中挖掘出频繁项集,针对数据 挖掘的经典关联规则 Apriori 算法的瓶颈问题提出了改进的方法。算法将事物数据库映射到布尔型数组中,然后所有的操 作都针对数组元素值展开。这样大大减少了数据库的扫描次数。算法利用数组的随机访问特性及布尔型数据的简单 “与”操作 ,直接产生频繁项集 ,而不产生大量的候选项集。经理论分析和实验结果显示该算法在效率上明显优于 Apriori 算法。 关键词 :数据挖掘 ;关联规则 ;Apriori 算法 ;频繁项集 中图分类号 : TP311 文献标识码 :A 文章编号 :1673 - 629 X( 2010) 06 - 0084 - 05 An Improved Apriori Algorithm f or Min ing Assoc iat ion Rules ZHAN G Guang2lu1 ,L E I J ing2sheng2 ,WU Xing2hui1 (1 . School of Mat hematics and Statistics , Hainan Normal U niversity , Hai kou 571158 ,China ; 2 . School of Informatio n Science and Technology ,Nanjing U niversity of Post s and Teleco mmunicatio ns ,Nanjing 211815 ,China) Abstract :Associatio n rule mining is an impo rtant part of research co ntent in data mining. In o rder to efficiently and quickly mine all f re2 quent itemset f ro m t he t ransactio n database ,an imp roved algo rit hm of mining associatio n rules is p resented fo r t he bot tleneck p ro blem of t he classic Ap rio ri algo rit hm. The t ransactio n database is mapped to Bool array , t hen all t he operatio ns are carried o ut based o n array ele2 ment s value , t hereby reducing t he database scanning f requency. Then use bit wise“AND”operatio n and rando m access characteristics of array , a direct co nsequence of f requent itemset s , rat her t han have a large number of candidate set s , t hereby imp roving t he efficiency of t he algo rit hm. Key words :data mining ;associatio n rules ;Ap rio ri algo rit hm ;f requent itemset p redeter mined minimum support count , min - sup . Sec2 o nd , generating interesting associatio n rules f ro m t he f re2 quent itemset s , i . e . , t hese rules must satisfy minimum support and minimum co nfidence . Because t he seco nd step is much less costly t han t he first step , t he overall perfor2 mance of mining associatio n rules is deter mi