纵向数据部分线性模型的基于高斯伪似然的高效估计

The Estimator Based on Gaussian Pseudo-likelihood for Partial Linear Models with Longitudinal Data

在纵向数据部分线性模型中,基于高斯伪似然构造相关阵的相合正定估计,进而得到回归参数的高效估计,给出了估计的算法及基于高斯伪似然的相关阵模型的选择准则.模拟表明,基于高斯伪似然方法得到的参数估计是高效的.

This paper propose the consistent and positive definite correlation matrix based on a Gaussian pseudo-likelihood method and obtain an efficient estimator of the regression parameters for partial linear models with longitudinal data. The estimation algorithm and the method for Gaussian pseudo-likelihood based selection of working covariance models were then given. Simulations show that the parameter estimation based on the Gauss pseudo-likelihood method is efficient.