Weight Composite Quantile Regression

Through the idea of quantile regression was first introdused in1760, though produced early, but the complexity when using the quantile regression brought in is a great chan-llenge. Nowdays quantile regression have already achieved numerical calculation through statistical software such as R-language and Stata software and by specific procedure editor. With the optimizing computing method, the working efficiency about quantile regression continually enhance. This progress made it can be applied to and developed in various domains.In this paper, we propose a new improved estimation method-Weight Composite Quantile Regression, based on the Quantile Regression. First, we illustrate the idea of the method. Then use this method for hereroscedasticity non-parametric regression model to estimate, provide estimation results and its asymptotic property, the description of weight solving. This thesis can be summarized brifly as follows:In the second chapter, we analyze the heteroscedasticity non-parametric regression model and illustrate its Weight Composite Quantile Regression estimation and deduc-tion.In the third chapter, according to the method of the Weight Composite Quantile Regression, we demonstrate the numerical simulation results of some examples. Further-emore, we analyza the robustness of the method.At last, we provide the theoretical proof of asymptotic property for the use of the Weight Composite Quantile Regression.