计算机与数字工程2018，Vol.46Issue(5)：992-996,5.DOI:10.3969/j.issn.1672-9722.2018.05.029
贝叶斯网络参数学习中的连续变量离散化方法研究
Research on Discretization Methods of Continuous Variables of Parameter Learning in Bayesian Network
刘晓明 1李盼池 1刘显德 1肖红1
作者信息
- 1. 东北石油大学计算机与信息技术学院 大庆163318
- 折叠
摘要
Abstract
It can qualitatively and quantitatively analyze the dependencies between attributes,and do probabilistic reasoning. In parameter learning of Bayesian network,it is usually assumed that all variables are discrete or continuous variables obeying Gauss-ian distribution,so it is necessary to do the discretization for those variables in reality which disobey the assumptions.In this paper, two different methods(equal width,ChiMerge)are used to discrete data set,then the results of discretization are used to construct the corresponding Bayesian network and do parameter learning in Netica.Finally,the resulting Bayesian network is used to do some simple forecast and analysis.关键词
贝叶斯网络/连续变量/离散化Key words
Bayesian network/continuous variables/discretization分类
军事科技引用本文复制引用
刘晓明,李盼池,刘显德,肖红..贝叶斯网络参数学习中的连续变量离散化方法研究[J].计算机与数字工程,2018,46(5):992-996,5.
