Statistical inference of a class of parametric varying coefficients in a regional quantile regression model

In order to solve the statistical inference problem of a class of parametric varying coefficients regional quantile regression, we propose an effective estimation of model parameters based on the idea of a weighted composite quantile regression. A random weighted resampling method was used to construct a rejection domain for significance tests of parameters in the model given limited samples. The numerical simulation results indicate that the proposed test statistic can effectively screen out covariates in the model. Finally, we applied the method to overseas study data and analyzed the impact of various factors on the chance of admission at different quantiles, in order to select the factors that have a significant impact on the chance of admission.