论文标题
对分类数据的进一步推断 - 贝叶斯方法
Further Inference on Categorical Data -- A Bayesian Approach
论文作者
论文摘要
本文讨论了与二维分类数据相关的三个不同的推论问题。考虑使用对称和不对称的超级参数的共轭先验分布。新构想的不对称先验基于类别的感知偏好。使用相关矩阵已经显示了通过引入参数之间测量关联的概念来扩展独立性测试。使用封闭形式的集成，蒙特卡洛集成和MCMC方法从后验分布中估算了不同参数组合的概率，以从分类数据中获取进一步的推断。贝叶斯计算是使用R编程语言完成的，并使用适当的数据集进行了说明。研究强调了利用潜在参数的分布形式的贝叶斯推论的应用。
Three different inferential problems related to a two dimensional categorical data from a Bayesian perspective have been discussed in this article. Conjugate prior distribution with symmetric and asymmetric hyper parameters are considered. Newly conceived asymmetric prior is based on perceived preferences of categories. An extension of test of independence by introducing a notion of measuring association between the parameters has been shown using correlation matrix. Probabilities of different parametric combinations have been estimated from the posterior distribution using closed form integration, Monte-Carlo integration and MCMC methods to draw further inference from categorical data. Bayesian computation is done using R programming language and illustrated with appropriate data sets. Study has highlighted the application of Bayesian inference exploiting the distributional form of underlying parameters.
