Coefficient Model And The Semi-parametric Generalized Linear Model Statistical Analysis

Varying-cofficient model and semi-parametric model,useful in many problems especially for Spatial data ,Biometric and Econometric applications,become an important research field in the regression .This paper presents the varying-cofficient model and semi-parametric generalized linear model ,especially for the research and application of the diagnostics and the generalized weighted least squares estimation of the varying-cofficient model.In the second chaper we study the statistics diagnostics of varying-cofficient generalized linear model , introduce the usual diagnostics models , get the concise diagnostics expressions based on case deletion or subject deletion models , establish an equivalence between the case deletion model and mean-shift outlier model from which we derive tests for outliers, establish a test of hypothesis on the base of mean-shift model , get score statistic of outliers, and the first order aproximation statistics. We study the diagnostics based on deviance. We also discuss the weighted least squares estimation, the generalized weighted least squares estimation in varying-cofficient linear model and get cook distance.In the third chapter, we get the parameter using cross-validation and generalized cross-validation,develop the likelihood distance,generalized leverage and deviance in the semi-parametric generalized linear model, get residuals in semi-parametric generalized linear model and the equivalence of the deviance based on case deletion or mean-shift outlier models.