Mei Ling Huang
University of Windsor, Natural Sciences and Engineering Research Council of Canada
![IconGems-Print](images/IconGems-Print.png)
240 – Computationally Intensive Methods for Estimation and Inference
An Efficient Method for Quantile Regression
Mei Ling Huang
University of Windsor, Natural Sciences and Engineering Research Council of Canada
Ramona Rat
Brock University
Wai Kong Yuen
Brock University
Quantile regression has wide applications in many fields. Studies of heavy tailed distributions have also rapidly developed. Estimation of high conditional quantiles for multivariate heavy tailed distributions is an important and interesting problem. This paper proposes a new weighted quantile regression method in high quantile regression. The Monte Carlo simulations of the bivariate Pareto distribution show good efficiency of the proposed weighted estimator relative to the regular quantile regression estimator. The paper also investigates a real-world example by using the proposed weighted method. Comparisons of the proposed method with existing methods are given.