Finite sample properties of adaptive regression estimators
通过大量蒙特卡洛模拟，研究了线性回归模型中自适应M估计和L估计的有限样本性质，并给出了选择修整比例、得分函数估计量等参数的指导建议。
Abstract The finite sample properties of adaptive M- and L-estimators for the linear regression model are studied through extensive Monte Carlo simulations. We provide guidelines for choosing the trimming proportion and estimating the score function for adaptive L-estimators. Suggestions are also given on choosing the trimming parameters, the score function estimators, the initial n-consistent estimator and estimating the Fisher information in the construction of adaptive M-estimators. We also investigate the effect of design dimension on the various adaptive estimators and the accuracy of the estimated standard errors. Keywords: Adaptive EstimatorsIterative Adaptive EstimatorsRobust EstimatorsScore FunctionFinite Sample