Sequential Diagrams and Influence Diagrams: A Complementary Relationship for Modeling and Solving Decision Problems
Zvi Covaliu · 1996
Abstract The conventional influence diagram paradigm is deficient in representing and solving asymmetric decision problems, which abound in the practice of decision analysis. A sequential diagram is a directed graph that, while sharing the node set of an influence diagram, captures asymmetric and time aspects of sequential decision problems as effectively as a decision tree. We define reduction operations on sequential diagrams and focus on the close relationship between sequential diagrams and influence diagrams in extensive form. We propose that the concurrent reduction of the two graphs and the complementary representation of a problem at all stages of solution provide insight into a problem’s structure not available by any of the graphs alone, streamline computations, and may indicate potential inadequacies in the initial model.