一种基于进制转换的数值型关联规则分析方法  

A Method for Quantitative Association Rule Mining Based on Decimal-binary Conversion

作　　者：刘均[1] 李人厚[1] 郑庆华[1] 

机构地区：[1]西安交通大学电信学院,西安710049

出　　处：《计算机工程》2005年第9期17-18,33,共3页Computer Engineering

基　　金：国家自然科学基金资助项目(60373105)

摘　　要：针对基于区间划分的数值型关联规则分析方法存在的问题,提出了一种基于进制转换的分析方法。此方法将十进制的属性值转换为二进制,并将二进制数值的每一位映射为二个布尔型属性,在此基础上,再采用布尔型关联规则分析方法。相对于基于区间划分的方法,此方法的优点是转化过程不存在失真,且能有效地降低后续分析过程的时间复杂度。Aiming at the issues on the method of quantitative association rule mining based on section partition,this paper introduces a method based on decimal-binary conversion. In the method, all numeric values are converted into the binary values, and then each digit of binary values is mapped to two Boolean attributes, the method of BAR mining is adopted for analyzing the association rules between the Boolean attributes. The advantages of this method is that no distortion is introduced during type conversion of source data, and the complexity of sequent BAR mining can be reduced efficiently.

关 键 词：数值型关联规则 布尔型关联规则 完整模式 进制转换 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]