Online Program Home
My Program

Abstract Details

Activity Number: 187 - Contributed Poster Presentations: Section on Nonparametric Statistics
Type: Contributed
Date/Time: Monday, July 29, 2019 : 10:30 AM to 12:20 PM
Sponsor: Section on Nonparametric Statistics
Abstract #305159
Title: Model-Based Quantile Regression: Analyzing Excess-Zero Response
Author(s): Erika Cunningham*
Companies: Duke University
Keywords: joint quantile regression; censoring; excess-zero; semi-continuous; nonparametric regression; Tobit regression

Quantile regression is a useful tool in understanding complex predictor-response relationships. In 2017, Yang and Tokdar presented a method for joint quantile regression across all quantile levels, opening the way for rigorous analysis of whether and how predictor effects vary between levels. We demonstrate how their method offers a comprehensive and model-based regression analysis framework. We illustrate how to interpret coefficients, improve and compare models, and obtain predictions under this framework. In an ecological application, we analyze how red maple abundance relates to site covariates. A complete absence of the species contributes excess zeros, which we treat as left censored in the spirit of a Tobit regression. By utilizing the generative nature of the joint quantile regression model, we are additionally able to make inference about the probability of the response being zero.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2019 program