Deriving inventory-control policies with genetic programming

Abstract.

One of the key areas of operations and supply chain management is inventory control. Inventory control determines which quantity of a product should be ordered when to achieve some objective, such as minimizing cost. Inventory-control policies are typically derived analytically, and this requires advanced mathematical skills and can be quite time consuming. In this paper, we present an alternative approach for solving inventory-control problems that is based on Genetic Programming. Genetic Programming is an optimization method that applies the principles of natural evolution to optimization problems. One of the key characteristics of Genetic Programming is that it does not require the specification of how a problem should be solved, but only the specification of what needs to be solved. After the user has specified the problem, GP searches for a solution without significant human involvement. The solutions generated by GP can be simple algorithms or closed-form expressions that represent the decision variables, i.e., the order point and the order quantity as a function of the problem parameters. However, expert knowledge in inventory control is still essential for building the inventory models and determining the parameters of Genetic Programming. Genetic Programming searches for both the structure and the parameters of the optimal solution. For simple settings, the structure and the parameters of the optimal solution can be found. For complex settings, near-optimal solutions that outperform traditional heuristics can be found if the structure of the optimal solution is known.