基于频繁2-项集的数量关联规则挖掘方法研究  

Study of Quantitative Association Rule Mining Based on Frequent 2-itemsets

作　　者：游晋峰[1,2] 冯山[1,2] 

机构地区：[1]四川师范大学计算机软件实验室,四川成都610066 [2]四川师范大学数学与软件科学学院,四川成都610066

出　　处：《四川师范大学学报（自然科学版）》2011年第1期128-133,共6页Journal of Sichuan Normal University（Natural Science）

基　　金：四川省教育厅自然科学重点基金(2003A079)资助项目

摘　　要：关联规则挖掘方法自提出以来已有很多改进算法,但均局限于布尔关联规则的挖掘.已有的数量关联规则挖掘主要考虑了连续属性值离散化、最优的数量关联规则挖掘等问题,但存在过小支持度和过小置信度问题.研究了这一问题并提出了一个在频繁2-项集的基础上挖掘数量关联规则的改进算法.它不仅可以用于典型的购物篮分析,还可以用于购物篮分析不能完成的关联规则挖掘问题,如带数量的捆绑销售问题.Since the idea of association rule mining was put forward,many improved algorithms have been provided,but they are limited in the mining of the Boolean association rules.Most of the association rule mining based on quantitative relations are mainly focused on discretization of continuous attributes and optimized quantitative association rule mining.But there are two problems about minimum support and minimum confidence.In this paper,a new algorithm of quantitative association rule mining based on frequent 2-itemsets is provided.It can be applied not only for typical basket analysis,but also for those mining problems that the present basket analysis algorithm is not able to achieve,such as bundled sales problem with quantitative.

关 键 词：频繁2-项集 APRIORI算法 数量关联规则 布尔关联规则 数据挖掘 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]