粗糙集理论的连续属性自适应离散化算法  

An Adaptive Discretization Algorithm for Continuous Attributes in Rough Set Theory

作　　者：杨素敏 陈翔 刘启明 崔静 YANG Su-min;Chen Xiang;Liu Qi-ming;Cui-jing(State Key Laboratory of Complex Electromagnetic Environment Effects on Electronics and Information System,Luoyang Henan 471003,China;Army Engineering University,Shijiazhuang Hebei 050003,China)

机构地区：[1]电子信息系统复杂电磁环境效应国家重点实验室,河南洛阳471003 [2]陆军工程大学石家庄校区,河北石家庄050003

出　　处：《计算机仿真》2023年第5期404-407,共4页Computer Simulation

基　　金：国家重点实验室项目:电子信息系统复杂电磁环境效应国家重点实验室课题(CEMEE2021Z0103B)。

摘　　要：针对多个连续属性离散时较少考虑属性重要性对离散质量的影响,以及离散造成的数据不相容问题,提出了粗糙集理论的自适应连续属性离散算法,算法利用粗糙集正域计算各属性的重要度,按照属性重要度有低到高的顺序对各属性依次进行离散,并在保证信息系统不可分辨关系不变的情况下进行断点的选取。在减少离散点总数的情况下,算法解决了不相容问题和较多地保留了重要属性的特征点。与CAIM离散算法和布尔运算的粗糙集离散算法从离散断点数、算法相容性和算法计算复杂度等几个方面的仿真比较结果表明了所提算法的有效性。An adaptive continuous attribute discretization algorithm based on rough set theory is proposed to address the impact of attribute importance on discretization quality and the problem of data incompatibility caused by discretization when discretizing multiple continuous attributes.Firstly,importance degree of each attribute was calculated,each attribute was discretized in ascending order of the importance degree,and redundant breakpoints were removed as much as possible with ensuring that the indiscernibility of information system remains unchanged.While reducing the total number of discrete points,the algorithm solves the problem of incompatibility and retains the characteristic points of important attributes.Compared with the CAIM algorithm and the Boolean algorithm in terms of the number of discrete breakpoints,algorithm compatibility and computational complexity,the experimental results show the effectiveness of the proposed algorithm.

关 键 词：连续属性 粗糙集理论 离散化算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]