摘要
粗糙集理论是一种有效的处理不一致、不精确和不完备等各种信息的数学分析工具。属性约简算法是粗糙集理论的关键技术之一,是数据挖掘研究的一个重要课题,也是知识获取中研究的关键问题之一。高效的属性约简算法使属性约简的求解被证实是一个NP-Hard问题,它通常是一个预处理阶段,使适应决策表上的分类分析。本文提出一种有效的方法——SEGMENT-SIG,可以得到最小约简子集,保持决策表的分类一致性。本文对算法最坏的时间计算复杂度进行了分析,该算法的输出是两种不同的分类器,一个是IF-THEN规则体系,另一个是决策树。
Rough set theory is an effective mathematical analysis approach to process inconsistent, uncertain and incomplete data and so on. Attribute reduction algorithm is one of the key technologies of rough set theory, and it is one of the important issues in data mining field, it is also a key problem of knowledge acquisition. High-efficiency attribute reduction is proved a NP-Hard prob- lem. It is generally regarded as a preprocessing phase, it also adapts to classificatory analysis of decision tables. This paper puts forward a new method SEGMENT-SIG, the method can find a minimal attribute subset, which preserves classificatory consis- tency of a decision table. Its worst-ease computational complexity is analyzed. The output of the algorithm are two different kinds of classifiers. One is an IF-THEN rule system, the other is a decision tree.
出处
《计算机与现代化》
2013年第8期68-72,77,共6页
Computer and Modernization
基金
黑龙江省自然科学基金资助项目(F201201)
关键词
粗糙集理论
属性约简
决策表
rough set theory
attribute reduction
decision table
