重尾数据的混合Erlang分布拟合方法

PH分布具有良好的解析性质,重尾数据的PH近似是随机模型分析中的重要课题.混合Erlang分布是一种常见的PH分布,本文利用EM方法给出了重尾数据的混合Erlang分布拟合算法,通过Matlab软件对两类常见重尾数据进行实验,结果令人满意.最后,文章对混合Erlang分布拟合算法的进一步改进做了总结.     

一类混合分红策略下的广义Erlang(n)风险模型

本文考虑混合分红策略下索赔来到间隔为广义Erlang(n)分布的更新风险模型,利用指数分布的无记忆性,分别得到破产前期望折现分红函数和折现分红的矩母函数满足的积分-微分方程及其边界条件.最后给出索赔为指数分布及索赔来到间隔为广义Erlang(2)分布的风险模型的期望折现分红函数的精确表达式.     

Erlang(2)扩散风险模型的Gerber-Shiu期望折现惩罚函数

将由布朗运动刻画的随机干扰项加入到Erlang(2)风险模型中,在模型中引入了由Gerber和Shiu定义的期望折现惩罚函数,并给出了这类模型的Gerber-Shiu函数所满足的积分微分方程.     

偏正态混合模型的惩罚极大似然估计

在分析具有异质性和非对称性数据时,偏正态混合模型提供一种比经典的Gauss混合模型更为灵活的建模方式.然而,由于无界的似然函数和发散的形状参数,该模型的极大似然估计并未被正确定义,进一步导致不理想的推断过程.为同时解决这两个问题,本文基于惩罚似然提出一种新的估计方案,并证明在混合分布的类别个数大于或等于真实的类别个数时,相应的惩罚极大似然估计是强相合的.同时,本文也提出相应的惩罚EM (expectation maximization)算法来计算惩罚估计.最后,通过模拟分析与现有方法比较研究估计方法在有限样本下的表现,并采用两个实例说明方法的有效性.     

一类零膨胀半参数混合效应模型的惩罚似然估计

本文将半参数线性混合效应模型推广应用到一类具有零膨胀的纵向数据或集群数据的研究中,提出了一类新的半参数混合效应模型,然后利用广义交叉核实法选取光滑参数,通过最大惩罚似然函数方法与EM算法给出了模型参数部分与非参数部分的估计方法,最后,通过模拟和实例说明了本文方法的有效性.     

零膨胀泊松混合回归中的变量选择

来源于不同总体的数据异质性较大,数据“零取值”较多且离散度大,可利用零膨胀泊松(ZIP)混合回归模型建模分析,然而混合模型中自变量较多.为了筛选出重要变量,本文利用自适应LASSO对ZIP混合回归模型进行变量选择,即在似然函数中加入惩罚项,再利用EM算法估计参数.通过模拟,验证了该方法在变量选择和参数估计中的有效性.同时,将ZIP混合回归模型应用于预测借贷失败次数的实际数据分析,筛选出对借贷失败有重要影响的因素.最后,通过比较各模型的预测效果,得到ZIP混合回归模型优于泊松(Poisson),负二项(NB)和ZIP回归模型.     

一类Sparre Andersen风险模型的Gerber-Shiu函数

本文考虑了一类相邻两次索赔的时间间隔服从Erlang($n$)和Erlang($m$)的混合分布的Sparre Andersen风险模型.主要目的是研究Gerber-Shiu函数$\phi_\delta(u)$,首先证明了$\phi_\delta(u)$满足一个高阶的积分微分方程,然后讨论了广义Lundberg方程根的性质,在此基础上导出了$\phi_\delta(u)$的拉普拉斯变换并且证明了$\phi_\delta(u)$满足一个更新方程,最后给出了一个例子.     

多维Erlang混合的次序统计量及其在多生命理论中的应用（英文）

本文研究来自多维Erlang混合分布的具有相关性的随机变量的次序统计量并将相关结论应用到多生命理论中.利用熟知的数学归纳法推导出次序统计量分布密度函数的解析表达式并证明其分布为一维Erlang混合分布,寿险精算理论中常见的重要统计量可以算出精确解.这一结论将随机变量相互独立情形下的结果推广到变量存在相关性的情形.     

Sparre Andersen风险模型的破产问题(英)

本文主要研究了一类Sparre Andersen模型,其索赔时间间隔的分布为指数分布与Erlang(n) 分布的混合.得到了当初始资金u趋于无穷大时,破产概率ψ(u)的确切表达式和渐近表达式.     

一类具有随机保费风险模型的破产问题

本文考虑了一个保费收入过程为复合Poisson过程,且索赔时间间隔分布为广义Erlang(n)分布的风险模型,给出了其罚金折现期望函数所满足的瑕疵更新方程以及渐近表达式和精确表达式.     

Ruin Probability for Finite Erlang Mixture Claims Via Recurrence Sequences

Methodology and Computing in Applied Probability - A new procedure to find the ultimate ruin probability in the Cramér-Lundberg risk model is presented for claims with a mixture of m Erlang...     

条件Erlang分布双参数加法定理的推广

X(m)和Y(k)服从参数(m,λ)和(k,μ)的Erlang分布且相互独立.证明了在X(m)相似文献   

Moments of the Time of Ruin, Surplus Before Ruin and the Deficit at Ruin in the Erlang（N） Risk Process

In this paper we consider the ＂penalty＂ function in the Erlang（n） risk model. Using the integro- differential equation we established, we obtain the explicit expressions for the moments of Erlang（2） risk model. When the claim size distribution is Light-Tailed and the penalty function is bounded, we obtain the exact representations for the moments of Erlang（n） risk model.     

Evaluation of (R,Q)-policies for two-level inventory systems with generally distributed customer inter-arrival times

In this paper we consider a two-level inventory system with a central warehouse and a number of retailers. All facilities apply continuous review (R,Q)-policies. We first extend Forsberg's exact Poisson model to the case with unit demand and customer inter-arrival times that are Erlang distributed. In the case with generally distributed customer inter-arrival times we approximate by Erlang distributions. We use two different methods to choose the approximate Erlang distribution. The first method means that we, for each retailer, choose the Erlang distribution that has the exact mean and minimum difference in standard deviation. Our second method means that we, for each retailer, choose the Erlang distribution of customer inter-arrival times that gives the exact mean and minimum difference in the standard deviation of the demand per unit of time instead of the inter-arrival time. Both methods are tested on 38 simulated cases. In all cases both methods give the same approximation.     

The expectation of aggregate discounted dividends for a Sparre Anderson risk process perturbed by diffusion

In this paper, we study the expectation of aggregate dividends until ruin for a Sparre Andersen risk process perturbed by diffusion under a threshold strategy, in which claim waiting times have a common generalized Erlang(n) distribution. For this strategy, we assume that if the surplus is above certain threshold level before ruin, dividends are continuously paid at a constant rate that does not exceed the premium rate, and if not, no dividends are paid. We obtain some integro-differential equations satisfied by the expected discounted dividends, and further its renewal equations. Finally, applying these results to the Erlang(2) risk model perturbed by diffusion, where claims have a common exponential distributions, we give some explicit expressions and numerical analysis. Copyright © 2007 John Wiley & Sons, Ltd.     

带息力的Erlang(2)风险过程下的一类积分方程

本文考虑了带息力的Erlang(2)风险模型，利用Sundt和Teugels(1995)，Yang和Zhang(2001a，2001b和2001c)文中的技巧，得到了生存概率所满足的积分方程和指数型的积分方程，然后研究了生存概率的Laplace-Stieltjes变换所满足的二阶微分方程．     

A class of Sparre Andersen risk process

In this paper, we investigate a renewal risk model in which the distribution of the interclaim times is a mixture of two Erlang distributions. First, the Laplace transform and the defective renewal equation for the Gerber-Shiu function are derived. Then, two asymptotic results for the Laplace transform of the time of ruin are given when the initial surplus tends to infinity for the light-tailed claims and heavy-tailed claims, respectively. Finally, an explicit expression for the Gerber-Shiu function is given.