Mathematical Modelling of Decision-Making: Application to Investment
Atefeh Hasan-Zadeh · Advances in Decision Sciences · 2019
In this paper, the mathematical modelling of decision-making problems based on probabilistic graphical models is presented. The models, in addition to random variables, also include decision and profit variables. The proposed decision models contain one or more decisions and their purpose is to assist the decision maker in choosing the best decision under conditions of uncertainty. The techniques used include modelling and evaluating decision trees, as well as evaluating, modelling and presenting the algorithm of influence diagrams (as the expansion of the Bayesian networks) to show the communication structure of the problem and thus provide a coherent presentation with an effective evaluation. The limited memory influence diagram and dynamic decision networks (as a dynamic Bayesian network) have also been developed to avoid the limitations of the influence diagram limitation. An application of the exposed models is presented in investment decision-making.