Statistical Inference In Varying-coefficient Partially Nonlinear Models With Missing Data

Regression models are often used in many fields such as medicine,biology,economics,management science,industry,agriculture and engineering design.The semi-parametric regression model has strong adaptability,and the regression functions can be estimated by using the observed values.To deal with complex data in different kinds of fields,many important semiparametric regression models have been proposed and developed.In this thesis,we mainly consider the estimation of interest parameter for varying-coefficient partially nonlinear model with missing data.For the varying-coefficient partially nonlinear model with response variables missing at random,we investigate the estimation of both parametric and nonparametric components.In order to deal with missing data,logistic parametric models are applied to estimate missing probability,which not only avoids the curse of dimensionality,but also is easy to calculate.In this thesis,we obtained the estimators of parametric and nonparametric components by inverse probability weighted least-squares method.The resulting estimators are shown to be asymptotically normal.The two estimators given in the second part are compared with the method of ignoring missing values under different missing probabilities and different finite samples through simulations,and the estimator which is obtained by using the inverse probability weighted least-squares estimator performed better.Finally,we analyse the actual data.