参数估计和变量选择的二次推断函数方法研究新进展

New Progress in Study of Quadratic Inference Function Method for Parameter Estimation and Variable Selection

二次推断函数已成为纵向数据分析的重要工具。文章介绍了参数和半参数模型的二次推断函数方法研究成果及最新进展,主要包括边际模型和混合效应模型中基于二次推断函数的参数估计、变量选择;二次推断函数方法在经验似然、测量误差模型方面的研究成果。指出了二次推断函数方法尚待解决的一些问题并对其未来的研究进行展望。

Quadratic inference function has become an important tool for longitudinal data analysis. This paper introduces the research findings and latest research progress of quadratic inference functions(QIF) method for parametric and semi-parametric models,mainly including parameter estimation,variable selection based on the quadratic inference functions in marginal model and mixed effect model, and the research achievements of quadratic inference function method in empirical likelihood and measurement error model. The paper also points out some problems to be solved in quadratic inference function method and prospects its future research.