基于最小二乘近似的模型平均方法

本文提出基于最小二乘近似的模型平均方法.该方法可用于线性模型、广义线性模型和分位数回归等各种常用模型.特别地,经典的Mallows模型平均方法是该方法的特例.现存的模型平均文献中,渐近分布的证明一般需要局部误设定假设,所得的极限分布的形式也比较复杂.本文将在不使用局部误设定假设的情形下证明该方法的渐近正态性.另外,本文...

Model averaging by least squares approximation

In this paper, we propose a model averaging method by adopting least squares approximation;as a result, the method can be used in linear regression models, generalized linear models, quantile regression, and so on. In particular, the classic Mallows model averaging is a special case of the proposed model averaging method.Unlike the existing literature on the distribution of model averaging where a local misspecification assumption is utilized and the resulting limiting distribution is complicated, we prove the asymptotic normality of the proposed estimator in the current paper without using the local misspecification assumption. In addition, the proposed method is extended to a high-dimensional situation. Numerical results show the promise of the proposed method.