基于t分布下混合联合位置与尺度模型的参数估计

t分布是分析厚尾数据的重要统计工具,本文基于t分布提出了稳健的混合联合位置和尺度参数的回归模型,通过EM算法给出该模型参数的极大似然估计,通过随机模拟试验验证了所提出方法的有效性.本文结合实际数据验证了该模型和方法具有实用性和可行性.     

缺失偏t正态数据下线性回归模型的统计推断

本文研究缺失偏t正态数据下线性回归模型的参数估计问题,针对缺失偏t正态数据,为使样本分布更加接近真实分布,改善模型的回归系数、尺度参数、偏度参数和自由度参数的估计效果,提高参数估计的稳定性,提出一种适合缺失偏t正态数据下线性回归模型的修正随机回归插补方法.通过随机模拟和实例研究,同随机回归插补,多重随机回归插补方法比较,结果表明所提出的修正随机回归插补方法是有效可行的.     

混合非线性联合均值与方差模型的统计推断

在异质总体中,混合回归模型是最重要的统计数据分析工具之一.提出了混合非线性联合均值与方差模型,通过EM算法研究了该模型参数的极大似然估计,并通过随机模拟实验验证了所提出方法的有效性.最后,结合实际数据验证了该模型和方法具有实用性和可行性.     

改进的强相关数据的变量选择方法

针对高维强相关数据的变量选择问题,本文提出了改进的变量选择方法.该方法先利用自适应弹性网方法(Aenet)在原始的强相关数据上建立模型,选出对响应变量起重要作用的群组变量和独立变量;再通过偏最小二乘方法(PLS)对选出的变量作模型估计;最后,将两种方法得到的估计系数做线性组合,并以此系数来建立回归模型.新模型具有精度高、解释性好的优点,数值实验验证了该方法的有效性.     

基于偏Laplace正态数据下位置、均值回归模型的参数估计

参数估计是一种基本的统计推断形式,也是统计学的一个重要分支.在分析偏态数据时,我们比较关注数据的众数、中位数和均值,但是偏Laplace正态数据的众数和中位数难以精确求出,因此用位置参数来近似代替.故本文提出偏Laplace正态数据下位置和均值回归模型,并研究该模型的参数估计,模拟和实例研究结果表明本文提出的模型和方法是科学合理的.     

尖峰厚尾保险损失数据的统计建模

保险损失数据的一个重要特点是尖峰厚尾性,即既有大量的小额损失,又有少量的高额损失,使得通常的损失分布模型很难拟合此类数据,从而出现了对各种损失分布模型进行改进的尝试.改进后的模型一方面要有较高的峰度,另一方面又要有较厚的尾部.最近几年文献中出现的改进模型主要是组合模型,即把一个具有非零众数的模型(如对数正态分布或威布尔分布)与一个厚尾分布模型(如帕累托分布或广义帕累托分布)进行组合.讨论了这些组合模型的性质和特点,并与偏t正态分布和偏t分布进行了比较分析,最后应用MCMC方法估计模型参数,并通过一个实际损失数据的拟合分析,表明偏t分布对尖峰厚尾损失数据的拟合要优于目前已经提出的各种组合模型.     

长度偏差右删失数据剩余寿命的分位数回归

本文研究长度偏差数据下剩余寿命分位数模型的估计方法,充分考虑有偏抽样机制对模型估计的影响.如果忽略这种有偏性会导致估计产生严重偏差甚至错误的结果.本文首先针对长度偏差右删失数据的剩余寿命分位数提出了对数形式的线性回归模型,对删失变量与协变量独立和不独立的两种情况利用估计方程给出了模型参数的估计.其次,通过经验过程和弱收敛理论给出了参数估计的相合性和渐近正态性.最后,本文对提出的估计方法进行了数值模拟并用该方法对奥斯卡奖数据进行分析.     

偏正态混合效应模型中方差分量函数的参数Bootstrap推断

偏正态混合效应模型通过引入偏度参数,以便更好地刻画实际数据偏态特征,所以被广泛应用于众多实际领域.进一步,方差分量的假设检验一直是该模型的热点研究问题.因此,有必要在偏正态分布下系统讨论混合效应模型中方差分量函数的统计推断问题.首先,分别基于参数Bootstrap方法和广义方法探讨单个方差分量、方差分量之和、方差分量之比的单边假设检验和区间估计问题.其次,Monte Carlo结果表明,在所给样本量和参数设置下,参数Bootstrap方法大多数情况下优于广义方法.最后,将上述方法应用于空气质量指数的案例研究中,以验证所给方法的合理性与有效性.     

基于Laplace分布下混合联合位置与尺度模型的参数估计

Laplace分布是分析厚尾数据的重要统计工具之一,本文基于Laplace分布提出了稳健的混合联合位置和尺度参数的回归模型,通过EM算法给出了该模型参数的极大似然估计,通过随机模拟试验验证了所提出方法的有效性.本文结合实际数据说明了该模型和方法具有实用性和可行性.     

高维纵向数据的模型平均估计

高维数据的模型选择是当今统计学研究的一个热点问题,但关于高维纵向数据方面的模型平均却少见研究,文章提出了一种利用删组交叉验证准则对高维纵向数据进行模型平均估计的方法,在最小化预测残差意义下,以删组交叉验证为准则,证明了其渐近最优性,并通过模拟研究表明,该模型平均方法在估计效果上要优于其它一些传统的模型选择和平均方法.     

Proportional Mean Residual Life Model with Varying Coefficients for Length-Biased and Right-Censored Data

Length-biased data are encountered in many fields,including economics,engineering and epidemiological cohort studies.There are two main challenges in the analysis of such data:the assumption of independent censoring is violated and the assumed model for the underlying population is no longer satisfied for the observed data.In this paper,a proportional mean residual life varyingcoefficient model for length-biased data is considered and a local pseudo likelihood method is proposed for estimating the coefficient functions in the model.Asymptotic properties are investigated for the proposed estimators.The finite sample performance of the proposed methodology is demonstrated by simulation studies.Finally,the method is applied to a real data set concerning the Academy Awards.     

Super-efficiency in stochastic data envelopment analysis: An input relaxation approach

This paper addresses super-efficiency issue based on input relaxation model in stochastic data envelopment analysis. The proposed model is not limited to using the input amounts of evaluating DMU, and one can obtain a total ordering of units by using this method. The input relaxation super-efficiency model is developed in stochastic data envelopment analysis, and its deterministic equivalent, also, is derived which is a nonlinear program. Moreover, it is shown that the deterministic equivalent of the stochastic super-efficiency model can be converted to a quadratic program. As an empirical example, the proposed method is applied to the data of textile industry of China to rank efficient units. Finally, when allowable limits of data variations for evaluating DMU are permitted, the sensitivity analysis of the proposed model is discussed.     

Semiparametric inference for transformation models via empirical likelihood

Recent advances in the transformation model have made it possible to use this model for analyzing a variety of censored survival data. For inference on the regression parameters, there are semiparametric procedures based on the normal approximation. However, the accuracy of such procedures can be quite low when the censoring rate is heavy. In this paper, we apply an empirical likelihood ratio method and derive its limiting distribution via U-statistics. We obtain confidence regions for the regression parameters and compare the proposed method with the normal approximation based method in terms of coverage probability. The simulation results demonstrate that the proposed empirical likelihood method overcomes the under-coverage problem substantially and outperforms the normal approximation based method. The proposed method is illustrated with a real data example. Finally, our method can be applied to general U-statistic type estimating equations.     

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.     

Non-probabilistic Bayesian update method for model validation

Model validation is the principal strategy to evaluate the accuracy and reliability of computational simulations. A systematic model validation procedure including uncertainty quantification, model update and prediction is described based on a non-probabilistic interval model. The crucial technical challenge in model validation is limited data, thus the non-probabilistic interval model is adopted to describe uncertain parameters. To establish the model update formula, the concepts of the interval escape rate and interval coverage rate are first described. Then, not only can the possibility of failure be estimated but also the credibility of the possibility of failure based on the proposed model validation method. The data in the validation experiment are used to update the credibility of each interval model, while the data from the accreditation experiment are used to conduct a final check of the validated models. To demonstrate that the proposed method can be applied to model validation problems successfully, a validation benchmark, the static frame challenge problem, is implemented. In addition, a practical aviation structure engineering validation problem is described. The results of these two validation problems show the feasibility and effectiveness of the proposed model validation method. The theoretical framework proposed in this paper is also suitable for model validation of computational simulations in other research fields.     

Unsupervised quadratic surface support vector machine with application to credit risk assessment

Unsupervised classification is a highly important task of machine learning methods. Although achieving great success in supervised classification, support vector machine (SVM) is much less utilized to classify unlabeled data points, which also induces many drawbacks including sensitive to nonlinear kernels and random initializations, high computational cost, unsuitable for imbalanced datasets. In this paper, to utilize the advantages of SVM and overcome the drawbacks of SVM-based clustering methods, we propose a completely new two-stage unsupervised classification method with no initialization: a new unsupervised kernel-free quadratic surface SVM (QSSVM) model is proposed to avoid selecting kernels and related kernel parameters, then a golden-section algorithm is designed to generate the appropriate classifier for balanced and imbalanced data. By studying certain properties of proposed model, a convergent decomposition algorithm is developed to implement this non-covex QSSVM model effectively and efficiently (in terms of computational cost). Numerical tests on artificial and public benchmark data indicate that the proposed unsupervised QSSVM method outperforms well-known clustering methods (including SVM-based and other state-of-the-art methods), particularly in terms of classification accuracy. Moreover, we extend and apply the proposed method to credit risk assessment by incorporating the T-test based feature weights. The promising numerical results on benchmark personal credit data and real-world corporate credit data strongly demonstrate the effectiveness, efficiency and interpretability of proposed method, as well as indicate its significant potential in certain real-world applications.     

A novel recursive subspace identification approach of closed-loop systems

In this paper, a subspace model identification method under closed-loop experimental condition is presented which can be implemented to recursively identify and update the system model. The projected matrices play an important role in this identification scheme which can be obtained by the projection of the input and output data onto the space of exogenous inputs and recursively updated through sliding window technique. The propagator type method in array signal processing is then applied to calculate the subspace spanned by the column vectors of the extended observability matrix without singular value decomposition. The speed of convergence of the proposed method is mainly dependent on the number of block Hankel matrix rows and the initialization accuracy of the projected data matrices. The proposed method is feasible for the closed-loop system contaminated with coloured noises. Two numerical examples show the effectiveness of the proposed algorithm.     

A novel multi-information fusion grey model and its application in wear trend prediction of wind turbines

The small and fluctuating samples of lubricating oil data render the wear trend prediction a challenging task in operation and maintenance management of wind turbine gearboxes. To deal with this problem, this paper puts forward a method to enhance the prediction accuracy and robustness of the grey prediction model by introducing multi-source information into traditional grey models. Multi-source information is applied by creating a mapping sequence according to the sequence to be predicted. The significance of the key parameters in the proposed model was investigated by numerical experiments. Based on the results from the numerical experiments, the effectiveness of the proposed method was demonstrated using lubricating oil data captured from industrial wind turbine gearboxes. A comparative analysis was also conducted with a number of selected other models to illustrate the superiority of the proposed model in dealing with small and fluctuating data. Prediction results show that the proposed model is able to relax the quasi-smooth requirement of data sequence and is much more robust in comparison to exponential regression, linear regression and non-equidistance GM(1,1) models.     

A Midpoint-Radius approach to regression with interval data

In this paper, a revisited interval approach for linear regression is proposed. In this context, according to the Midpoint-Radius (MR) representation, the uncertainty attached to the set-valued model can be decoupled from its trend. The estimated interval model is built from interval input-output data with the objective of covering all available data. The constrained optimization problem is addressed using a linear programming approach in which a new criterion is proposed for representing the global uncertainty of the interval model. The potential of the proposed method is illustrated by simulation examples.     

滑动数据重心预测方法及其应用研究

本文在研究多因素数据重心法的基础上,进一步提出滑动数据重心预测方法,该方法是对原始样本数据提出了一种新的数据处理方法,大大降低了由于历史数据组中的异常点对预测结果产生的破坏性。通过建立我国钢材消费量与国内生产总值(GDP)的计量动态模型对该方法与多因素数据重心预测法进行对比研究。同时利用时间序列自回归AR(p)对计量动态模型的初级预测结果进行差值校正,并将该方法应用于我国2015年、2020年的钢材消费量预测。对比研究表明该方法使得预测结果更加精确、稳健。