中文摘要 本论文主要分析讨论了线性回归模型中 Bayes估计方法的优良性,并将它与其他估计方法做了对比分析。在线性回归模型中的均方误差准则下得到Bayes 估计比起最小二乘估计有更好的精确度;其优良性还可以通过比较各自的风险函数的大小来做出评价。由于先验分布的引入,最大后验估计方法比起极大似然估计的确有优势;由于损失函数的不同应该相应的采取不同的 Bayes 估计方法;而在线性回归模型中回归系数的条件期望估计值的均方误差比最小二乘估计的均方误差更小。6643
关键词 Bayes估计 线性回归 先验分布 损失函数
毕业设计说明书(论文)外文摘要
Title The Bayes estimator of the linear regression model
Abstract
The thesis is mainly concerned with the optimality of Bayesian
estimation method of the linear regression model and compared with other
estimation method to analyse. Under the MSEM of the linear regression model
we can also find the Bayesian estimator is more acurate than the LSE. In
the optimality we can make an evaluation by comparing of the size of their
risk functions; The maximum posterior density estimator which compared
with the maximum likelihood estimator does have advantages because of the
prior distribution.As the loss functions are different we should use the
different Bayes estimation methods.And the conditional expectation
estimator is less than the LSE about the regression coefficient under the
MSEM of the linear regression model.
Keywords Bayes estimator;Linear regression;Prior distribution;Loss function
目 次