青岛大学学报（自然科学版）2017，Vol.30Issue(1)：9-13,5.DOI:10.3969/j.issn.1006-1037.2017.02.03
图对数线性模型性质的研究
Research on the Properties of the Graphical Log-linear Models
曲敏 1许成 1蒋文超1
作者信息
- 1. 青岛大学数学与统计学院,青岛 266071
- 折叠
摘要
Abstract
By combining the Markov network with the log-linear models in contingency tables, the graphical log-linear models are established and the maximum likelihood estimate is studied.With the decomposition of the graphical models according to the maximum clique, the maximum likelihood estimate can be obtained.Thestructure and independence conditions of the five-dimensional graphical log-linear models are analyzed, and its formula of maximum likelihood estimation are obtained.关键词
图对数线性模型/马尔科夫网/独立性/极大似然估计Key words
graphical log-linear models/Markov network/independence/maximum likelihood estimate分类
数理科学引用本文复制引用
曲敏,许成,蒋文超..图对数线性模型性质的研究[J].青岛大学学报（自然科学版）,2017,30(1):9-13,5.
