缺失数据下局部M-估计的相合性和渐近正态性

在缺失响应变量的不完全数据下,对非参数回归模型进行研究,利用稳健的变窗宽局部线性回归的方法,给出了回归函数m(x)的估计,并证明了局部M-估计具有相合性和渐近正态性.所提出的方法继承了局部多项式回归的优点并且克服了最小二乘方法缺乏稳健性的缺点.并且使用变窗宽提高了所得M-估计的可塑性,使之能成功地处理空间非齐次曲线、异方差性及非均匀设计密度.所得估计的渐近结果为求渐近最优方案以及直接从数据估计最优变窗宽提供了理论基础.     

回归函数的小波支持向量机鲁棒估计法

小波网络具有小波的多尺度特性和神经网络的自学习功能,在回归估计中得到广泛的应用,但其性能受到样本中粗差的严重影响.虽然以M-估计作为目标函数可以解决这个问题,但由于其对应的影响函数由残差绝对值决定,因此如何选择初始参数值成为一个关键问题.为此,提出回归函数的小波支持向量机鲁棒估计方法(小波支持向量回归,WSVR,Wavelet Support Vector Regression).该方法中首先提出并证明了一种新的小波支持向量机(WSVM,Wavelet Support Vector Machine),用于确定初始参数值方法,这种方法能够确定合理的网络结构和合适的初始参数值,保证含有粗差的样本点的残差绝对值较大;然后使用一种构造的M-估计作为目标函数,并提出了自适应确定阈值方法.仿真结果表明,使用这种方法得到的回归模型不仅具有良好的多尺度逼近特性,而且有较好的鲁棒性和较高的推广性能,具有较高的理论和应用价值.     

变系数模型变窗宽和一步局部M-估计

用变窗宽和一步局部M-估计对变系数模型的系数参数进行估计,得到了估计的渐近正态性.     

����~���Ƶ�ָ����ط����Ի��ЧӦģ�͵����Լ���

Homogeneity of variance and correlation coefficients is one of assumptions in the analysis of longitudinal data.However, the assumption can be challenged. In this paper, we mainly propose and analyze nonlinear mixed effects models for longitudinal data with exponential correlation covariance structure, intend to introduce Huber's function in the log likelihood function and get robust estimation (M-estimation) by Fisher scoring method. Score test statistics for homogeneity of variance and correlation coefficient based on M-estimation are then studied. A simulation study is carried to assess the performance of test statistics and the method we proposed in the paper is illustrated by an actual data example.     

递推M-估计的强收敛性

本文对u1(t)为单调递增这一比较常见的情形和在其它一些比较常见的条件 下,证明了M-估计递推算法中回归系数和散布矩阵的强相合性.     

非随机设计矩阵递推M-估计的强收敛性

对解释变量为非随机设计矩阵xi这一比较常见情形的M-估计递推算法进行了研究．在一些比较常见的条件下，证明了M-估计递推算法中回归系数和散布矩阵的强相合性．     

多参数M估计与随机加权法

本文考虑了标准化的多参数Ｍ估计之误差分布的随机加权逼近，得到的逼近精度为ｏ（ｎ￣（－１／２）），从而我们推广了［１］和［２］的结果。     

Strong consistency of M-estimates in linear models

The strong consistency of M-estimates of the regression coefficients in a linear model under some mild conditions is established, which is an essential improvement over the relevant results in the literature on the moment condition. Especially, in some important circumstances, onlyE|ψ(e_k)|^q for some q > 1 is needed, where ψ{e_k} is some score function of random error.     

General relative error criterion and M-estimation

Relative error rather than the error itself is of the main interest in many practical applications. Criteria based on minimizing the sum of absolute relative errors (MRE) and the sum of squared relative errors (RLS) were proposed in the different areas. Motivated by K. Chen et al.’s recent work [J. Amer. Statist. Assoc., 2010, 105: 1104-1112] on the least absolute relative error (LARE) estimation for the accelerated failure time (AFT) model, in this paper, we establish the connection between relative error estimators and the M-estimation in the linear model. This connection allows us to deduce the asymptotic properties of many relative error estimators (e.g., LARE) by the well-developed M-estimation theories. On the other hand, the asymptotic properties of some important estimators (e.g., MRE and RLS) cannot be established directly. In this paper, we propose a general relative error criterion (GREC) for estimating the unknown parameter in the AFT model. Then we develop the approaches to deal with the asymptotic normalities forM-estimators with differentiable loss functions on ? or ?\{0} in the linear model. The simulation studies are conducted to evaluate the performance of the proposed estimates for the different scenarios. Illustration with a real data example is also provided.     

变系数模型变窗宽局部M-估计的相合性

用变窗宽局部M-估计对变系数模型的系数函数进行估计,得到了估计的相合性和渐近正态性.所采用的方法继承了局部多项式回归的优点并且克服了最小二乘方法缺乏稳健性的缺点.变窗宽的使用提高了局部M-估计的可塑性,并使得它们能成功地处理空间非齐性曲线、异方差性及非均匀设计密度.