多项分布无信息先验的探究

Non-information Prior for the Multinomial Distribution Research

贝叶斯方法是对多项分布参数进行点估计的重要方法.在统计学家对无信息先验作了大量的研究工作的基础上,通过推导得到一组关于该无信息先验分布的结论并利用这组结论研究其后验均值.结果表明:后验均值可使后验均方差达到最小.

Bayesian method is important for the estimation of the multinomial parameters. Statisticians have made a great number of experiments about the research of a non-informative prior. This thesis introduces this prior and discusses its estimation.