Metric based complexity reduction for multivariate extremes

For the extremal analysis of multivariate data, the (empirical) tail correlation matrix is an important characteristic that measures extremal dependence. In this project, we further explore the properties of this matrix and its connection to max-stable representations of the underlying dependence structure. Via the embedding of the correlation matrix into a distance function and subsequent simplification of this distance function in the form of line and tree metrics we detect extremal dependence patterns.