摘要
离散化是Rough集理论研究的一个重要内容,目前基于Rough集的离散化算法很难做到高效率和高识别率兼顾.文中分析了基于断点重要性算法和基于属性重要性算法的特点,确定了离散化思路,提出了一种基于Rough集的集成离散化算法.该算法能够有效降低候选断点的数目,快速地实现决策表的离散化.实验结果表明,文中算法保持了与已有算法可比的识别率,且运行效率更高.
Discretization is one of the important points of Rough sets theory study.The discretiztion algorithm based on Rough sets can hardly have both high efficiency and high recognition rate.The features of algorithm based on breakpoints and attribute are analyzed and the integrated discretiztion algorithm based on rough sets is offered,which can reduce the number of candidate breakpoints and quickly realizes the discretization of decision table.The experiments show that this algorithm keeps high recognition rate and has high working efficiency.
出处
《重庆三峡学院学报》
2010年第3期59-63,共5页
Journal of Chongqing Three Gorges University
关键词
粗糙集
决策表
连续值属性
集成离散化
Rough sets
decision table
continuously valued attributes
integrated discretization
