|
Activity Number:
|
494
- Simulation-Based Approaches
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 6, 2020 : 10:00 AM to 2:00 PM
|
|
Sponsor:
|
Section on Statistical Computing
|
|
Abstract #309753
|
|
|
Title:
|
On Nonparametric Extreme Quantile Regression
|
|
Author(s):
|
Percy Brill and Andrew Kenig* and Mei Ling Huang
|
|
Companies:
|
Brock University and University of Windsor and Brock University
|
|
Keywords:
|
Conditional quantile;
extreme value distribution;
Frèchet distribution;
generalized Pareto distribution;
Hill estimator;
nonparametric regression
|
|
Abstract:
|
Quantile regression (QR) estimates conditional quantiles with wide applications in the real world. Estimating extreme conditional quantiles is an important and difficult problem. The regular quantile regression method often sets a linear model with estimating the coefficients to obtain the estimated conditional quantile. This approach may be restricted by the model setting, there are also computational difficulties. To overcome this problem, this paper proposes a two-stage nonparametric quantile regression method with a 6-step algorithm by using extrapolation. Monte Carlo simulations show good efficiency for the proposed nonparametric QR extrapolation estimator relative to the regular linear QR extrapolation estimator. The paper also investigates an Alberta Wildfire example by using the proposed method. Comparisons of the proposed method and existing methods are given.
|
Authors who are presenting talks have a * after their name.
