基于Legendre多项式的财产损失分布建模研究

在财产保险中,关于损失分布建模的问题,大部分研究都是着眼于传统的参数统计方法,其基本流程为:获取数据→选择参数模型→估计模型参数→指出拟合效果。在模型的选择过程中,人们一般都会假定总体服从几种常见的分布函数,如Pareto分布、Gamma分布、Log-     

POT模型在巨灾损失预测中的应用——基于MCMC方法的估计

极值统计学主要研究随机事件极端情况的统计规律性。运用POT模型拟合中国暴雨损失数据,确定损失超出量的分布形式。实证分析表明,借助POT模型对巨灾风险损失分布进行估计是较为合理的,但当数据量较小时,使用基于Gibbs抽样的MCMC方法估计POT模型的参数,可以解决样本数据不足导致的极大似然估计中误差增大的问题。     

非寿险精算中的数据尾部拟合与保费厘定

文章讨论了极值分布对非寿险精算中损失数据尾部的拟合和保费厘定方法,并进行了实例计算。研究表明:必须对应用极值分布的条件进行检验;对门限值确定的三种方法中自适应选择算法是较好方法;广义帕累托分布参数MLE估计能得到比较精确的估计结果。文章还给出了非寿险损失的超赔再保险纯保费的计算方法。     

半参数混合模型下的贝叶斯方法与应用

极端值估计是损失评估的重要研究部分,文章在贝叶斯方法的基础上,用半参数混合模型来拟合损失.在确定模型参数的过程中,运用贝叶斯方法对参数建模,将参数转化成随机变量,并基于马尔卡夫蒙特卡罗(MCMC)抽样得到参数的估计值.该方法的特点是参数数量少,通过抽样把参数转化成随机变量,给出所有参数可能取值的频率分布图.实证结果表明模型结果既考虑了参数的不确定性,又兼顾了损失的厚尾性.     

定时截尾样本下广义逆指数分布参数的Bayes估计

文章在定时截尾样本下,讨论了广义逆指数分布形状参数、可靠度和危险率的极大似然估计。基于指数先验分布,在熵损失、平方损失和Linex损失函数下分别得到形状参数、可靠度和危险率的Bayes估计,并给出了确定超参数的方法。利用数值模拟计算了估计量的各种估计均值和均方误差，研究结果表明,形状参数在熵损失和Linex损失函数下的估计精度较高;可靠度的Bayes估计整体优于极大似然估计;危险率的Bayes估计在Linex损失函数下的效果较好。     

关于我国地震灾害损失分布函数的研究

文章以1978～2006年间我国发生的183起地震灾害事故为样本,选取地震损失额、每年发生地震次数为指标,建立我国地震灾害损失分布函数。针对地震损失额利用经验剩余函数值分析损失分布集中程度、初步估计损失分布函数,分组处理样本数据、拟合分布图像、参数估计、单个样本非参数检验,确定损失分布函数为对数正态分布;针对每年发生地震灾害次数初步估计分布函数,通过历史频率与理论概率比较判断拟合效果、确定分布函数为泊松分布。     

基于C-K模型及其改进的高高龄人口死亡率模型研究

目前,国际上关于高龄人口死亡率估计的方法主要包括经典的参数模型估计方法,如用Gompertz分布、Makeham分布拟合高龄人口的死亡率。文章将着重探究Coale-Kisker模型在高高龄人口死亡率拟合中的应用,并针对高高龄人口死亡数据足够多时,对经典的Coale-Kisker模型提出适当的改进。     

逐次定数截尾下Pareto分布的可靠性分析

基于逐次定数截尾模型,文章选取未知参数的先验分布为无信息先验分布,分别在平方损失和LINEX损失下,讨论了Pareto分布的形状参数,失效率以及可靠度函数的Bayes估计。最后运用Monte Carlo方法对Bayes估计和极大似然估计的MSE,进行了模拟比较。结果表明在LINEX损失下的Bayes估计更优。     

本期导读

在非寿险业务中,对损失数据所服从的分布的精确估计是一个十分重要的问题.<非寿险损失分布建模的一般性方法>一文,利用平均超出函数、极大似然估计等方法系统地分析了损失分布的模型识别、参数估计和模型拟和检验的技术方法,并通过实例验证了在有大量损失数据情况下,利用计算技术解决非寿险损失分布模型拟和是一种有效的方法.     

非参数随机条件持续期模型及其迭代算法

 随机条件持续期（SCD）模型能有效刻画超高频时间序列中持续期的变化,但该模型假定期望持续期生成机制固定,且模型参数估计存在一定的困难。文章在不假定条件均值形式和冲击项分布的基础上结合核估计方法提出了非参数SCD模型及其迭代求解方法。然后,基于TEACD(1,1)模型生成的模拟数据,将非参数SCD模型与用卡尔漫滤波进行伪似然估计的参数SCD模型和用Gibbs抽样进行马尔科夫蒙特卡罗估计的参数SCD模型的拟合效果进行比较,实证表明在大样本条件下非参数SCD模型的拟合效果与用MCMC估计的参数SCD模型的拟合结果相差不大,但明显优于用QML估计的参数SCD模型的拟合结果,且非参数SCD模型能为参数SCD模型的参数设定提供参考。     

基于正态分布点值化的区间主成分评价法及应用

针对区间数据点值化过程中所存在的“代表性不足”的缺陷,提出了基于正态分布的点值化方法并将之应用于区间主成分评价法。通过与基于中心点值化的区间主成分法的比较,得到三个主要结论：第一,基于正态分布的点值化方法能将各样品的点值化结果导向指标均值,而非区间值的中心点;第二,基于正态分布的点值化结果增加了数据信息量;第三,基于正态分布点值化的区间主成分评价法提高了数据降维效果,具有更好的因子命名能力。应用结果表明,在考虑正态分布情况下,对区间数据的点值化处理方法具有较好的效果,基于正态分布点值化的方法可推广至基于区间数的评价和决策问题。     

Extreme value index estimator using maximum likelihood and moment estimation

ABSTRACT

When a distribution function is in the max domain of attraction of an extreme value distribution, its tail can be well approximated by a generalized Pareto distribution. Based on this fact we use a moment estimation idea to propose an adapted maximum likelihood estimator for the extreme value index, which can be understood as a combination of the maximum likelihood estimation and moment estimation. Under certain regularity conditions, we derive the asymptotic normality of the new estimator and investigate its finite sample behavior by comparing with several classical or competitive estimators. A simulation study shows that the new estimator is competitive with other estimators in view of average bias, average MSE, and coefficient of variance of the new device for the optimal selection of the threshold.     

An extension of the half-normal distribution

In this paper we introduce a new distribution, namely, the slashed half-normal distribution and it can be seen as an extension of the half-normal distribution. It is shown that the resulting distribution has more kurtosis than the ordinary half-normal distribution. Moments and some properties are derived for the new distribution. Moment estimators and maximum likelihood estimators can computed using numerical procedures. Results of two real data application are reported where model fitting is implemented by using maximum likelihood estimation. The applications illustrate the better performance of the new distribution.     

Generalized exponential distribution: different method of estimations

Recently a new distribution, named as generalized exponential distribution has been introduced and studied quite extensively by the authors. Generalized exponential distribution can be used as an alternative to gamma or Weibull distribution in many situations. In a companion paper, the authors considered the maximum likelihood estimation of the different parameters of a generalized exponential distribution and discussed some of the testing of hypothesis problems. In this paper we mainly consider five other estimation procedures and compare their performances through numerical simulations.     

Gamma-Maxwell distribution

A new two-parameter distribution, the gamma-Maxwell distribution, isproposed and studied. We generate the new distribution using the gamma-G generator of distributions. The proposal distribution can be seen as an extension of the Maxwell distribution with more flexibility in terms of the distribution asymmetry and kurtosis. We study some probability properties, discuss maximum-likelihood estimation and present a real data application indicating that the new distribution can improve the ordinary Maxwell distribution in fitting real data.     

New flexible models generated by gamma random variables for lifetime modeling

In this paper we introduce a new three-parameter exponential-type distribution. The new distribution is quite flexible and can be used effectively in modeling survival data and reliability problems. It can have constant, decreasing, increasing, upside-down bathtub and bathtub-shaped hazard rate functions. It also generalizes some well-known distributions. We discuss maximum likelihood estimation of the model parameters for complete sample and for censored sample. Additionally, we formulate a new cure rate survival model by assuming that the number of competing causes of the event of interest has the Poisson distribution and the time to this event follows the proposed distribution. Maximum likelihood estimation of the model parameters of the new cure rate survival model is discussed for complete sample and censored sample. Two applications to real data are provided to illustrate the flexibility of the new model in practice.     

A new compounding life distribution: the Weibull–Poisson distribution

In this paper, a new compounding distribution, named the Weibull–Poisson distribution is introduced. The shape of failure rate function of the new compounding distribution is flexible, it can be decreasing, increasing, upside-down bathtub-shaped or unimodal. A comprehensive mathematical treatment of the proposed distribution and expressions of its density, cumulative distribution function, survival function, failure rate function, the kth raw moment and quantiles are provided. Maximum likelihood method using EM algorithm is developed for parameter estimation. Asymptotic properties of the maximum likelihood estimates are discussed, and intensive simulation studies are conducted for evaluating the performance of parameter estimation. The use of the proposed distribution is illustrated with examples.     

A skew extension of the slash distribution via beta-normal distribution

In this work we introduce a generalization of the slash distribution using beta-normal distribution. This newly defined generalization is more flexible than the ordinary slash distribution and contains distributions that can be not only symmetric and unimodal, but also asymmetric and bimodal. We study the properties of the new generalized distribution and demonstrate its use on some real data sets considering maximum likelihood estimation procedure.     

A new weighted distribution as an extension of the generalized half-normal distribution with applications

In this study, a new extension of generalized half-normal (GHN) distribution is introduced. Since this new distribution can be viewed as weighted version of GHN distribution, it is called as weighted generalized half-normal (WGHN) distribution. It is shown that WGHN distribution can be observed as a single constrained and hidden truncation model. Therefore, the new distribution is more flexible than the GHN distribution. Some statistical properties of the WGHN distribution are studied, i.e. moments, cumulative distribution function, hazard rate function are derived. Furthermore, maximum likelihood estimation of the parameters is considered. Some real-life data sets taken from the literature are modelled using the WGHN distribution. It is seen that for these data sets the WGHN distribution provides better fitting than the GHN and slashed generalized half-normal (SGHN) distributions.     

Strong consistency of presmoothed Kaplan–Meier integrals when covariables are present

It is known that the Kaplan–Meier estimation may be improved via presmoothing methods. In this article, we introduce an extended presmoothed Kaplan–Meier estimator in the presence of covariates. The main result is the strong consistency of general empirical integrals based on such an estimator. As applications, one can obtain a consis-tent multivariate empirical distribution under censoring, and also can obtain a consistent estimation of regression parameters. We illustrate the new estimation methods through simulations and real data analysis.