Causality for Decision‐Making
Louis Anthony Cox · Wiley StatsRef: Statistics Reference Online · 2019
Abstract Rational decisions seek to cause preferred outcomes by selecting actions or policies that make them more likely. Decision optimization models therefore typically include causal models of how outcome probabilities depend on the decision maker's choices, as well as on other direct causes such as the uncontrollable state of nature – a catchall term for causes of outcomes that the decision maker cannot choose. These causal models are often left implicit in the objective functions (e.g., expected utility) that the decision maker seeks to maximize and in the constraints on allowed values of decision variables. Making them explicit through causal network models that display the dependencies (indicated by arrows) among decision variables and other uncertain quantities (represented as nodes in the network) allows a variety of network‐based algorithms to be used to solve for optimal decisions and invites natural extensions to include choices over time by multiple decision makers. This article reviews causal network models and optimization algorithms for single and multiple decision makers.