Using markov decision processes to solve business problems and an application

By the intensified competition through technological improvements and changing customer requirements, uncertainty in the decision-making processes of businesses has been increasing. A high level of uncertainty entails making more objective decisions while forming future projections, therefore businesses to utilize management science methods. In determining decisions involving uncertainty while making future plans, Markov Decision Process (MDP), one of the stochastic management science methods, could provide assistance to managers. It is realized that there are studies covering MDP applications oriented to several decisions of businesses in the literature.Various methods such as value iteration, policy iteration or linear programming (LP) can be used in the solution of MDP?s when only one objective such as profit maximization or cost minimization is considered. Compared to other methods, LP provides more flexibility by enabling the consideration of several system constraints. However the decisions made by business while operating in an intensive competition environment require considering multiple and usually conflicting objectives simultaneously and making decisions providing to approximate all of those objectives as far as possible. Goal programming (GP), one of the multi-objective decision-making techniques, can be used to solve such problems. In this study, the production/inventory system of a business operating in the automotive supplier industry is modeled as MDP and three goals of business are considered together by constructing GP model oriented to solve the problem. The aim of this study is to provide an integrated perspective involving the utilization of MDP and GP approaches together for the solution of stochastic multi-objective decision problems.