摘要
通过用Bayes方法对(a,b)类分布进行分析,研究相关方差与期望的关系,并给出a与b的矩估计和极大似然估计(MLE).在极大似然估计基础上,利用Lindley逼近引理,给出(a,b,0)类的Bayes估计,并运用MATLAB进行相关模拟.模拟结果表明,对于(a,b,0)类分布的估计,若样本数量较大,则选择Bayes估计更好;反之,选择矩估计更好.
By using Bayesian method to analyze a class of（a,b）distribution,we studied the relationship between variance and expectation,and gave the moments estimation and maximum likelihood estimation（MLE）of aand b.Base on the MLE,we gave the Bayesian estimation of the classes distribution（a,b,0）by using Lindley approximation lemma,and used MATLAB to carry on the related simulation.Simulation results show that the choice of the Bayesian estimation is better if the sample size of（a,b,0）distribution estimation is large;conversely,the moment estimation is better.
出处
《吉林大学学报（理学版）》
CAS
CSCD
北大核心
2016年第3期480-486,共7页
Journal of Jilin University:Science Edition
基金
国家自然科学基金(批准号:11271155)
关键词
类分布
BAYES估计
矩估计
极大似然估计
期望
方差
class distribution
Bayesian estimation
moment estimation
maximum likelihood estimation（MLE）
expectation
variance
