广义线性模型参数的自适应拟似然估计

对广义线性模型参数的一种拟似然估计的理论给予了彻底的处理. 在该估计中,响应变量的未知的协方差阵是通过样本去估计的.证明了所定义的估计量具有下述意义上的渐近有效性：当样本量n→∞时, 该估计有渐近正态性,且其极限分布的协方差阵重合于当响应变量的协方差阵完全已知时,拟似然估计的极限分布的协方差阵.     

几何分布截尾样本场合步进应力加速寿命试验TFR模型下的统计分析

讨论了几何分布产品在步进应力加速试验TFR模型下寿命分布.给出了其寿命分布函数步进形式,在截尾样本场合利用极大似然估计方法和拟矩估计方法求出了未知参数的点估计,最后利用计算机模拟考察了说明本文方法的可行性.     

Ⅰ型双删失下逆Rayleigh分布的统计分析

在Ⅰ型双删失样本下,用极大似然法得到了逆Rayleigh分布尺度参数估计的迭代公式.根据遗失信息原则计算出了Fisher信息矩阵,由极大似然估计的渐近正态性得到了参数的置信区间.取共轭先验分布,在平方损失函数下,求得了未知参数、可靠度函数的贝叶斯估计和参数的等尾置信区间.根据后验预测密度函数,得到了预测值的估计.通过Monte Carlo随机模拟,得到了多种估计值,并进行了比较,结果表明在小样本场合贝叶斯估计要优于极大似然估计.     

基于ARFIMA-GARCH模型的混成检验

本文研究了ARFIMA-GARCH模型的混成检验问题.基于拟极大指数似然估计,给出了平方残差自相关函数的渐近性,进而建立了基于平方残差自相关函数的混成检验统计量.通过实例分析,表明可利用基于平方残差自相关函数的混成检验统计量来诊断检验由拟极大指数似然估计方法拟合的ARFIMA-GARCH模型.     

半参数回归模型拟极大似然估计的渐近分布

用拟极大似然估计方法研究了误差为AR(1)时间序列的半参数回归模型,得到了参数及非参数的拟极大似然估计量,并研究了它们的渐近分布.     

自然联系函数下自适应设计广义线性模型的渐近性质(英文)

本文在一些弱的条件下,对自然联系函数和自适应设计下广义线性模型的极大拟似然估计渐近性进行研究,获得了极大拟似然估计的渐近存在性、弱相合性、收敛速度及渐近正态性.并通过蒙特卡罗数值模拟的方法对所得结果进行验证.     

拟似然非线性模型中MQLE的弱相合性的充分条件

拟似然非线性模型包括广义线性模型作为一个特殊情形.给出了拟似然非线性模型中极大拟似然估计的弱相合性的一些充分条件,其中矩的条件要弱于文献中极大拟似然估计的强相合性的条件.     

非线性模型中拟似然估计的若干性质

本文主要讨论拟得分函数在广义正则线性无偏函数类中的性质，并证明拟似然估计在泛似拟然估计类中的渐近最优性。     

序约束下ARCH(0,2)模型参数估计与检验

本文研究了平稳ARCH（0，2）模型未知参数α的极大似然估计及有序约束时α的极大似然估计的渐近性质，给出了参数序关系（α1≥α2）的检验方法，并得出了似然比检验统计量的渐近分布。用二次规划的算法，给出求各种情况下参数α的极大似然估计的数值算法。     

Pareto分布环境因子的估计及其应用

给出了Pareto分布环境因子的定义,讨论了在定数截尾样本下Pareto分布环境因子的极大似然估计和修正极大似然估计,并尝试把环境因子用于可靠性评估中.最后运用Monte Carlo方法对极大似然估计,修正极大似然估计和可靠性指标的均方误差(MSE),进行了模拟比较,结果表明修正极大似然估计优于极大似然估计且考虑环境因子的可靠性评估结果较好.     

Convergence rates of MLE in a partly linear model

This paper considers the estimation for a partly linear model with case 1 interval censored data. We assume that the error distribution belongs to a known family of scale distributions with an unknown scale parameter. The sieve maximum likelihood estimator (MLE) for the model’s parameter is shown to be strongly consistent, and the convergence rate of the estimator is obtained and discussed.     

Modified Maximum Likelihood Estimators Based on Ranked Set Samples

The maximum livelihood estimator (MLE) using a ranked set sample (RSS) usually has no closed expression because the maximum likelihood equation involves both hazard and inverse hazard functions, and may no longer be efficient when the judgment ranking is imperfect. In this paper, we consider a modified MLE (MMLE) using RSS for general parameters, which has the same expression as the MLE using a simple random sample (SRS), except that the SRS in the MLE is replaced by the RSS. The results show that, for the location parameter, the MMLE is always more efficient than the MLE using SRS, and for the scale parameter, the MMLE is at least as efficient as the MLE using SRS, when the same sample size is used. Under the perfect judgment ranking, numerical examples also show that the MMLE has good efficiency relative to the MLE based on RSS. When the judgment error is present, we conduct simulations to show that the MMLE is more robust than the MLE using RSS.     

多周期Probit模型中MLE的存在性

本文我们讨论了多周期Probit模型中MLE的存在性问题,给出了当协方差阵已知时,参数的MLE存在的充要条件;当协方差阵未知但具有序列结构时,参数的MLE存在的一个必要条件和一个充分条件.     

On properties of estimators in nonregular situations for Poisson processes

We consider the problem of parameter estimation by observations of an inhomogeneous Poisson process. It is well known that if regularity conditions are fulfilled, then the maximum likelihood and Bayesian estimators are consistent, asymptotically normal, and asymptotically efficient. These regularity conditions can be roughly presented as follows: (a) the intensity function of the observed process belongs to a known parametric family of functions, (b) the model is identifiable, (c) the Fisher information is a positive continuous function, (d) the intensity function is sufficiently smooth with respect to the unknown parameter, and (e) this parameter is an interior point of the interval. We are interested in properties of estimators for which these regularity conditions are not fulfilled. More precisely, we present a review of results which correspond to the rejection of these conditions one by one and show how properties of the MLE and Bayesian estimators change. The proofs of these results are essentially based on some general results by Ibragimov and Khasminskii. Bibliography: 9 titles.     

Maximum likelihood estimate for discontinuous parameter in stochastic hyperbolic systems

The purpose of this paper is to study the identification problem of a spatially varying discontinuous parameter in stochastic hyperbolic equations. In previous works, the consistency property of the maximum likelihood estimate (MLE) was explored and the generating algorithm for MLE proposed under the condition that an unknown parameter is in a sufficiently regular space with respect to spatial variables.In order to show the consistency property of the MLE for a discontinuous coefficient, we use the method of sieves, i.e. the admissible class of unknown parameters is projected into a finite-dimensional space. For hyperbolic systems, we cannot obtain a regularity property of the solution with respect to a parameter. So in this paper, the parabolic regularization technique is used. The convergence of the derived finite-dimensional MLE to the infinite-dimensional MLE is justified under some conditions.     

Reducing bias of the maximum likelihood estimator of shape parameter for the gamma Distribution

The gamma distribution is an important probability distribution in statistics. The maximum likelihood estimator (MLE) of its shape parameter is well known to be considerably biased, so that it has some modified versions. A new modified MLE of the shape for the gamma distribution is proposed in this paper, which is consistent, asymptotically normal and efficient. For finite-sample behavior, the new estimator improves the traditional MLE not only for reducing bias but also for gaining estimation efficiency significantly. In terms of estimation efficiency, it dominates other existing modified estimators.     

不完全信息下型截尾Weibull分布参数的极大似然估计及贝叶斯估计

给出了不完全信息下 型截尾weibull分布参数的极大似然估计、无信息先验Bayes估计及多层Bayes估计,并指出针对一些具体模型还可以通过随机模拟来比较其估计精度.     

The Doubly-Pareto Uniform Distribution with Applications in Uncertainty Analysis and Econometrics

In this paper we present a novel four parameter continuous univariate distribution that can be motivated from at least two approaches. The first one views the distribution as a generalization of the uniform one that allows for uncertainty specification at the vicinity of its bounds (gradually) represented via two Pareto tails. The second one is that of an asymmetric heavy-tailed, peaked distribution with an unbounded domain with the property that the location of the mode is not uniquely determined but rather is described by a uniform range. Properties of the distribution are described and a maximum likelihood estimation (MLE) procedure for the mode location and the Pareto tails parameters is presented. The procedure is illustrated by means of an i.i.d. sample of standardized log-differences of quarterly monthly US certificate deposit interest rates for the period 1964–2004. The sample is constructed utilizing the Auto-Regressive Conditional Heteroscedastic (ARCH) time series model devised by the Nobel Laureate R.F. Engle (1982).     

Minimax estimators for location vectors in elliptical distributions with unknown scale parameter and its application to variance reduction in simulation

In this paper, we give an ever wider and new class of minimax estimators for the location vector of an elliptical distribution (a scale mixture of normal densities) with an unknown scale parameter. The its application to variance reduction for Monte Carlo simulation when control variates are used is considered. The results obtained thus extend (i) Berger's result concerning minimax estimation of location vectors for scale mixtures of normal densities with known scale parameter and (ii) Strawderman's result on the estimation of the normal mean with common unknown variance.Research partially supported by National Science Foundation, Grant #DMS 8901922.     

On the consistency of the MLE for Ornstein–Uhlenbeck and other selfdecomposable processes

In this paper we give easy to verify conditions for the strong consistency of the maximum likelihood estimator (MLE) in the case when data is sampled from a parametric family of selfdecomposable distributions. The difficulty arises from the fact that standard conditions for the consistency of the MLE are based on the pdf, which, for most selfdecomposable distributions, is not available in a closed form. Instead, our conditions are based on properties of the Lévy triplet (i.e. the Lévy measure, the Gaussian part, and the shift) of the distribution. Further, we extend out results to certain selfdecomposable stochastic processes, and, in particular, we give conditions in the case when the data is sampled from a Lévy or an Ornstein–Uhlenbeck process.