针对概率统计中的半参数广义线性模型(SGLMS),利用了正交空间、核函数、正交计算方法及独立随机分布的概念和性质,研究了广义线性模型中未知参数的估计问题,给出了回归参数的矩估计方法及其相应的性质,给出了参数估计量为一致有效性和近似正态性的充分条件,并证明了参数估计量是一致有效性和近似正态性的结论。所给出的回归参数的矩估计方法推广了已有文献中要求参数完全已知的假设,因此该方法和结果更具有一般性和理论价值。
In this paper, semiparametric generalized linear models (SGLMs) in probability and statistics are considered. By utilizing the properties of orthogonal space, kernel function, orthogonal computation method and independent random distributed functions, a moment estimation method of parameters regression is proposed and some properties are given. Sufficient conditions to ensure parameter estimations to be consistent efficiency and asymptotic normality are proposed. It is proved that the proposed moment estimations are asymptotically consistent and normal. The proposed methods and results extend the existing results. Therefore, they have general significance and theoretical value.