Optimizing Order Decisions for Perishable Product Inventory Management: A Policy-Based Dynamic Programming Approach

Managing inventory for perishable products presents unique challenges due to limited shelf life, demand uncertainty, and the risk of spoilage. This paper proposes a policy-based Dynamic Programming (DP) approach to find the best policy that optimizes order decisions for perishable product inventory management. The model considers factors such as demand variability, changes in price, and product shelf life to maximize profit while considering waste.A set of 1000 simulation experiments, inspired by real-world perishable product retail scenarios, demonstrates that the Predictive Policy significantly outperforms traditional inventory methods. The results indicate that the Predictive Policy increases total profit by 27% compared to the Base Stock Policy and by 16.9% compared to the Order-Up-To Policy. Additionally, the Predictive Policy reduces expired stock by 69.6% compared to the Random Policy and by 47.5% compared to the Base Stock Policy, highlighting its efficiency in waste reduction. These findings emphasize the effectiveness of data-driven decision-making in improving supply chain profitability and sustainability for perishable goods.