A1704
Title: Dependence maps: A graphical tool for visualizing multivariate dependence
Authors: Ivan Medovikov - Brock University (Canada) [presenting]
Abstract: A graphical tool for visualizing complex relationships between the variables is demonstrated that is based on the difference of empirical copulas of the underlying variables that we call a "dependence map". We also present a version of the tool designed to assess comonotonicity of vectors, that is - to highlight vector dependence. We demonstrate the applicability of the tool by visually studying the dependence between hourly wholesale electricity prices and grid features such as load and weather in New York. We demonstrate how the tool can uncover instances of non-linear (e.g. tail) dependence.
