Non-Exchangeable Random Variables Modelling Using Copulas

Abstract eng:
Estimating joint distribution function of a random vector (X, Y ) can be split into estimation of the marginal distribution functions FX and FY, and of the copula CX,Y (in short, C) describing the dependence structure of (X, Y ). Then FX,Y (x, y) = C (FX(x), FY (y)). For the second step majority of existing methods concern the exchangeable random variable case when CX,Y = CY,X (symmetric copulas). Among few classes of asymmetric copulas, we focus on the class of Archimax copulas. To fit copulas to non-exchangeable random variables we first estimate the best fitting Archimedean copula generator, then the asymmetry degree transformed by the generator and finally we choose the appropriate dependence function from parametric classes using the least square approach. The application is implemented in R and demonstrated on hydrological data..

• Export as: BibTeX | MARC | MARCXML | DC | EndNote | NLM | RefWorks
• Add to your basket