Reorder Decision System Based on the Concept of the Order Risk Using Neural Networks

Abstract

Due to the development of the modern information technology, many companies share the real-time inventory information. Thus the reorder decision using the shared information becomes a major issue in the supply chain operation. However, traditional reorder decision policies do not utilize the shared information effectively, resulting in the poor performance in distribution supply chains. Moreover, typical assumption in the traditional reorder decision systems that the demand pattern follows a specific probabilistic distribution function limits practical application to real situations where such probabilistic distribution function is not easily defined. Thus, we develop a reorder decision system based on the concept of the order risk using neural networks. We train the neural networks to learn the optimal reorder pattern that can be found by analyzing the historical data based on the concept of the order risk. Simulation results show that the proposed system gives superior performance to the traditional reorder policies. Additionally, managerial implication is provided regarding the environmental characteristics where the performance of the proposed system is maximized.