部分线性变系数模型的加权混合几乎无偏岭估计

Weighted Mixed Almost Unbiased Ridge Estimation of Partial Linear Variable Coefficient Model

文章研究随机线性约束条件下部分线性变系数模型的参数估计问题。为了克服多重共线性,融合Profile最小二乘估计、几乎无偏岭估计和加权混合估计构造了回归模型参数分量新的加权混合几乎无偏岭估计,并在均方误差矩阵准则下给出新估计量优于加权混合估计和几乎无偏岭估计的充要条件,最后通过数值模拟验证了所提出估计量的有限样本性质。

This paper studies the parameter estimation of partial linear variable coefficient model under stochastic linear constraints.In order to overcome the multicollinearity,the paper fuses the Profile least square estimation,almost unbiased ridge estimation and weighted mixed estimation to construct a new weighted mixed almost unbiased ridge estimation for parametric components of the regression model,and then gives the necessary and sufficient conditions for the new estimator to be superior to the weighted mixed estimator and the almost unbiased ridge estimator under the mean square error matrix criterion.Finally,the paper verifies the finite sample properties of the proposed estimator by numerical simulation.