带有误差变量混合分布模型的估计

Estimates of Mixture Distribution Model with Measurement Errors

在许多实际问题中,存在一些不可直接观测的变量,对此统计学家们提出了反卷积和混合分布模型来解决这一变量的分布的估计问题.本文对这一问题采用bootstrap模拟方法得出分布函数的估计,并进一步建立该分布函数的非参bootstrap百分位区间.在数值试验中将我们的处理方式与传统的EM算法得到的分布估计和正态逼近区间作比较,数值结果表明用bootstrap模拟方法得到的准确度更好,数值效果更理想.

There are many unobservable variables in some practical fields, for this, deconvolution and mixture distribution has been developed and used widely. In this paper, we consider the estimation of a distribution function when observations from this distribution are contaminated by measurement error. The approach for using mixture distributions and bootstrap simulations is used to solve this prob＇.em, For two parts,distribution function and confidence interval, we show that our result is much better than Clifford,B. C.