Improve Quadratic Inference Function Estimation In Marginal Regression Models

The regression analysis of this paper is to study the relationship between a network data.In order to integrate the network topology into statistical inference,we propose a shrinkage quadratic inference function based on the idea of shrinkage estimation in simple marginal regression model and semi-parametric marginal model,which uses both prior structure and data-driven correlations among network nodes.The improved estimation method is simple in concept and fast in computation,and has appealing large-sample properties.Finally,the simulation experiments show that the improved estimator has a smaller mean square error.