Quantile Regression Theory In The Application Of Risk Analysis

In order to solve practical problems, people often build a linear regression model to study the impact of the change of independent variable on the dependent variable. Classical least squares regression model with its theoretical perfect, easy to calculate and is widely used in the data analysis. Least squares regression only measures the impact of the independent variable on the dependent variable distribution center,but can not fully reflect the impact of the various parts of the entire distribution, and least squares regression error distribution is requested to meet the normality, independence, and homogeneity of variance in the actual where this requirement is difficult to meet the problem. In1978, Koenker and Bassett propose quantile regression theory basis for the independent variable due to variable conditions quantile regression, so all quantile regression model is got. Relative to the classical least-squares regression, quantile regression not only have application conditions are relatively easy to meet, and,to some extent, reflects the information, in particular, a specific area of the data, such as the extreme position of the data of all data.This paper studies the quantile regression model concept, quantile regression model to establishment and the parameters of the method including quantile regression model estimates and inspection,and provide a theoretical basis for the application of quantile regression in practice; then in the study river hydrological risk problem of flow change and river flood control system in the closure, quantile regression model estimates and results forecast for the model parameters are got,while the quantile regression model is given; finally, quantile regression model calculated results is compared with the least squares regression models to show that quantile regression model can provide more information, especially the tail. Two practical problems of calculation results show that the quantile regression results reflect the random nature of the flow changes and hydrological risks more fully.