摘要
几何分布是寿命分布中一种重要的分布.在寿命数据处理时,经常会使用混合分布进行分析或拟合,在对混合分布的参数进行估计时,运用常规的矩估计法、极大似然估计法会比较困难.应用EM算法对混合几何分布的参数进行估计,得到了参数的估计迭代公式,并利用Matlab软件进行了数据模拟,从而说明了估计的可行性.
Geometric distribution is an important distribution in the life time distribution.In the data processing of the life time distribution, often using mixed distribution to analysis or fit, when we try to estimate the mixed distribution, using the moment estimation and maximum likelihood estimation will be difficult. In this paper, we employ the EM algorithm to obtain the iterative formula of parameter estimation, the simulation is made by Matlab software to evidence the feasibility of estimation.
引文
[1]Dempster A.P,Laird N.M,Rubin D.B.Maximum likelihood estimation from incomplete data via the EM algorithm[J].Journal of the Royal Statistical Society,1977,39(1):1-38.
[2]朱利平,卢一强,茆诗松.混合指数分布的参数估计[J].应用概率统计,2006,22(2):137-150.
[3]田玉柱,韩学峰,田茂再.混合删失样本下混合指数分布的估计(英文)[J].应用概率统计,2017,33(2):191-202.
[4]Tian Yuzhu,Zhu Qianqian,Tian Maozai.Estimation for mixed exponential distributions under type-Ⅱprogressively hybrid censored samples[J].Computational Statistics and Data Analysis,2015,89:85-96.
[5]张宝龙.有限混合分布模型参数估计的EM算法及模拟[D].宁夏大学,2015.
[6]陈文强.混合泊松分布模型的参数估计问题[D].西北大学,2009.
[7]木拉提.吐尔德,胡锡健.EM算法在删失数据分布和混合分布参数估计中的应用[J].统计与决策,2011(15):161-164.
[8]吴绍敏,程细玉.几何分布场合恒加应力寿命试验下的混合数据分析[J].华侨大学学报(自然科学版),1997,18(1):6-10.
[9]何朝兵,刘华文.随机截尾情形下几何分布的参数估计[J].湘潭大学自然科学学报,2013,35(1):29-32.
[10]茆诗松,王静龙,濮晓龙.高等数理统计(第二版)[M].北京:高等教育出版社,2006.