半参数回归模型的几乎无偏岭估计

ALMOST UNBIASED RIDGE ESTIMATORS IN A SEMIPARAMETRIC REGRESSION MODEL

提出了半参数回归模型的几乎无偏岭估计,并与岭估计进行了比较,在均方误差意义下,几乎无偏岭估计优于岭估计,然后讨论了有偏参数的选取问题,最后,用模拟算例和实际应用说明了几乎无偏岭估计的有效性和可行性。

A semiparametric regression model is considered, and the almost unbiased ridge estimators （AURR） of both parameters and nonparameters is introduced without a restrained design matrix. The AURR is compared with ridge estimator （RR） under a mean square error （MSE）, the result shows that the former excels the latter, and the optimal value for the biased parameter is established by minimum MSE. The validity and feasibility of AURR are demonstrated by a simulation example.