基于Rough集的启发式约简中启发式规则比较研究

摘要

It is rather common to take attribute importance as heuristic attribute reduction rule. The article chooses to briefly introduce a few commonly studied attribute importance heuristics rules, such as attribute dependency, discerptibility matrix frequency, comentropy and other rules. Rules and algorithms are implemented by programming. Summaries are made by comparison to classical data set computation. From computational results, a few fundamental conclusions are acquired about the influences of different heuristic rules upon attribution reduction.%采用属性的重要性作为启发式属性约简规则比较普遍.选择几种研究较多的属性重要性启发式规则,如属性依赖度、区分矩阵频率、信息熵等,进行简要介绍.通过编程实现规则和算法、采用经典数据集的运算比较作了汇总,从运算结果分析中获取了不同启发式规则对属性约简影响的几个基本结论.