A Symmetric Prior for Multinomial Probit Models
A Symmetric Prior for Multinomial Probit Models
复制标题
多项概率模型的对称先验
DOI：
10.1214/20-ba1233
复制
发表时间：
2019
期刊：
影响因子：
--
通讯作者：
P. R. Hahn
中科院分区：
文献类型：
--
作者：
Lane F Burgette;David Puelz;P. R. Hahn
来自广泛使用的贝叶斯多项概率模型的拟合概率可以强烈地依赖于基本类别的选择，该基本类别用于唯一地识别模型的参数。本文提出了一种新的识别策略，以及模型参数的相关先验分布，使得先验关于结果类别的重新标记是对称的。新的先验允许有效的Gibbs算法对秩亏协方差矩阵进行采样，而无需求助于Metropolis-Hastings更新。
Fitted probabilities from widely used Bayesian multinomial probit models can depend strongly on the choice of a base category, which is used to uniquely identify the parameters of the model. This paper proposes a novel identification strategy, and associated prior distribution for the model parameters, that renders the prior symmetric with respect to relabeling the outcome categories. The new prior permits an efficient Gibbs algorithm that samples rank-deficient covariance matrices without resorting to Metropolis-Hastings updates.