Contact:Elizna Huysamen
- 021 808 3244
Location: Dept Statistics, Van der Sterr building, Cnr Bosman & Victoria street, Entrance 5, Room 2053, Stellenbosch
Copulas
are multivariate distribution functions with standard uniformly distributed
marginal distributions. The key to copula theory is Sklar’s Theorem. It states
that a copula is a dependence function, which interconnects the univariate
marginal distribution functions of a random vector and thereby models its joint
distribution function. As a direct consequence, the individual probabilistic
behaviour of random variables and their inherent dependence can be modelled
separately.
The talk aims to provide a
basic introduction to copula theory focusing on how copula models can be used
for modelling the dependence structure in data at hand. Supported by several
data examples including applications in finance and biomedicine the class of
regular vine copulas, which allows flexible dependence modelling even in high
dimensions, will be introduced.