|
Activity Number:
|
113
- Statistical Computing in Modern Statistics
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 8, 2022 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistical Computing
|
|
Abstract #322542
|
|
|
Title:
|
General Gamma-Based Copulas with Applications in Tail Dependence
|
|
Author(s):
|
Matthew Arvanitis* and Barry C Arnold
|
|
Companies:
|
USDA Forest Products Laboratory and UC Riverside
|
|
Keywords:
|
copula;
gamma components;
likelihood-free
|
|
Abstract:
|
Tail dependence occurs when collections of random variables exhibit significant dependence at only certain extremes of their support. Gamma-based copulas are particularly useful for modeling these phenomena. In this talk we discuss the details of one of the most general (bivariate) families of these copulas and how manipulating the parameters can lead to different configurations of tail dependence, and even configurations of multiple tail dependencies. This family has no general density but can be easily simulated. Therefore, parameter estimation and other analyses must be performed using computationally intensive likelihood-free methods. Some of these will be demonstrated. We demonstrate the flexibility of this family by exhibiting several interesting sub-families and show its practical significance with real-data examples.
|
Authors who are presenting talks have a * after their name.
