Graphical models for stochastic processes
Vanessa Didelez · 2003
Abstract Graphical models have proven to be a valuable concept in various fields of multivariate data analysis, as underpinned by several contributions to this book. However, the pure definition of conditional independence graphs is not satisfactory for the representation of two closely related dependence concepts: causal and dynamic dependence - both sharing for instance the property of being asymmetric, as opposed to conditional dependence. In graphical models directed edges symbolize specific sets of conditional independence hypotheses.