基于复合分位数回归的部分线性模型平均估计

Jackknife Model Averaging for Partial Linear ModelsBased on Composite Quantile Regression

针对部分线性模型的参数与非参数估计问题,基于复合分位数回归提出了一种稳健的模型平均估计量.为了提高估计效率,采用B样条的方法拟合子模型中的非参数函数.数值模拟和仿真实验证明了所提出的估计量预测效果优良.

Aiming at the problem of parameter and non-parameter estimation for partial linear models,a robust model averaging estimator based on composite quantile regression is proposed.To improve the efficiency of estimation,B-splines are used for the non-parameter function of the sub-model.Simulation experiments and practical applications verify the superiority of the proposed method.