基于粗糙集的贝叶斯网络连续变量离散化算法  

Rough set-based algorithm for discretizing continuous variables of Bayesian network

作　　者：瞿英[1,2] 李萍[1] 吴祈宗[1] 贾建 

机构地区：[1]北京理工大学管理与经济学院,北京100081 [2]河北科技大学经济与管理学院,石家庄050081 [3]河北远东哈里斯通信有限公司市场部,石家庄050200

出　　处：《计算机应用》2009年第9期2491-2493,2508,共4页journal of Computer Applications

基　　金：河北省教育厅软科学基金资助项目(2007205);河北省科技厅科技支撑计划项目(09213509D)

摘　　要：分析了布尔逻辑和粗糙集理论相结合的离散化算法在贝叶斯网络建模中应用的局限性,提出基于断点值逐次扫描划分实例对的连续变量离散化算法,改变了原算法中基于断点区间划分实例对的计算方法,应用实例表明,该算法时间复杂度和空间复杂度较低,具有较强的实用性,更适用于贝叶斯网络中的连续数据离散化处理。Based on the analysis of the limitation that the discretization algorithm of Rough Set （RS） and Boolean reasoning approach did not work well in Bayesian network, a new algorithm was put forward to distinguish two samples by the value of candidate cuts, not by the intervals determined by two candidate cuts. The case study indicates that the improved algorithm can reduce preferably the space complexity and time complexity of the discretization. It is effective on discretizing continuous variables of Bayesian network.

关 键 词：粗糙集理论 连续变量离散化 候选断点 贝叶斯网络 信息表 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]