Quantile Regression Model And Its Application In Study On The Influencing Factors Of Medical Consumption

When solving the practical regression analysis,It is difficult to achieve the classical assumption of least squares estimation,so the result is not ideal.Since quantile regression was proposed,it gradually become a research hotspot because of its wide range of applications,and can extract different information according to different quantile.With the rapid development of computer technology,the realization of quantile regression theory becomes more and more easy.This paper firstly reviewed the development of quantile regression theory.Then we studied weighted composite quantile regression of the partial linear model,obtained the estimators of the parametric and the nonparametric part,established the asymptotic normality of the estimators,and gave a selection method of optimal weights.Then we transform weighted composite quantile regression model to a standard linear programming problem.The results obtained from different distributions of the error in Monte Carlo simulation showed that the proposed method has good performance in finite samples,provided a powerful proof for the theoretical model.Finally,this paper analyzed the influencing factors of medical consumption by using quantile regression and variable selection model.We found that results of quantile regression variable selection model and the significance of multiple linear regression model’s parametric estimation are consistent,so the results of variable selection are reasonable.Then we analyzed the effect and influence degree of medical consumption based on the results of variable selection.Among the factors of medical expenditure,personal income,age,education level,prevalence and regional factors have a greater impact,and are positive effects.The empirical results show that compared with the least squares estimation,quantile regression can provide a more comprehensive analysis and dig out more deeper information.And then based on the second chapter,this paper analyzed the influencing factors of medical consumption,and found that this method consider multiple quantiles’ information,could clearlyreflect the main factors affecting medical consumption and correctly predict the effect of the factors and the degree of influence,the results are more comprehensive.