Computer Engineering and Applications ›› 2008, Vol. 44 ›› Issue (14): 152-154.

• 数据库、信号与信息处理 • Previous Articles     Next Articles

Approach and Dephi programming for attributes discretization and reduction

FAN Min1,2,ZOU Ping2,ZHU Xing-dong1   

  1. 1.Department of System Science and Applied Mathematics,Kunming University of Science and Technology,Kunming 650093,China
    2.Faculty of Management and Economic,Kunming University of Science and Technology,Kunming 650093,China
  • Received:2007-01-06 Revised:2008-03-03 Online:2008-05-11 Published:2008-05-11
  • Contact: FAN Min


范 敏1,2,邹 平2 ,朱兴东1   

  1. 1.昆明理工大学 理学院 系统理论与应用数学系,昆明 650093
    2.昆明理工大学 管理与经济学院,昆明 650093
  • 通讯作者: 范 敏

Abstract: This paper mainly discusses the attributes discretization in Rough Set theory.It begins with the algorithm based on the significance of the attributes value.Then it uses the results of references paper [5],the undistributed class and the significance,meanwhile,a heuristic algorithm based on general core set for attribute reduction;finally,the experiment shows that it can get better effect and it also shows the reduction results of UCI database using this algorithm.

Key words: Rough Set theory, discretization, attribute reduction, discernibility matrix, undistributed class, a heuristic algorithm

摘要: 主要对Rough Set理论中的离散化问题进行了研究,提出了基于属性值重要性的离散化算法,利用文献[5]中不可分辩类的定义、属性重要性的讨论及属性约简的启发式算法,得到了一种离散化与属性约简相结合的启发式算法,并通过Delphi程序验证了该算法对UCI数据库的有效性。

关键词: 粗糙集理论, 离散化, 属性约简, 可辨识矩阵, 不可分辩类, 启发式算法