基于Bayes后验概率的自变量与异常点的同时识别

Simultaneous Variable Selection and Outlier Identification Based on Bayes Posterior Probability

本文基于自变量与异常点识别隐变量的联合Bayes后验概率,给出了自变量与异常点同时识别的一般方法,且利用Gibbs抽样降低了Bayes后验概率的计算复杂度。其次,针对多值序次数据模型自变量与异常点的同时识别展开详细讨论,给出了同时识别的具体过程。最后通过模拟算例展示了本文方法的有效性。

This article gives the method of simultaneous variable selection and outlier identification based on the posterior probability of the variable and outlier latent indicator variables, and uses the Gibbs sampling to alleviate the compution of Bayes posterior probability. Secondly, we discuss the method of simultaneous identification for ordinal data model variable and outlier in detail, and give the process of simultaneous identification. Finally, a simulated example is used to show the efficiency of our method.