Files
Abbadi_Tahir_64252400_65222400_2025.pdf
Closed access - Adobe PDF
- 2.35 MB
Details
- Supervisors
- Faculty
- Degree label
- Abstract
- This thesis explores the ongoing challenges of managing inventory in supply chains affected by shifting demand and supply conditions. It reflects on the limits of conventional approaches like EOQ, JIT, and MRP, and gradually shifts focus to artificial intelligence techniques, especially reinforcement learning, as a way to handle complexity with more flexibility. To examine this in practice, a Q-learning algorithm was developed using real inventory data from a retail setting. Two models were tested: one based on stock levels alone, and another that also took weekly demand into account. The experiments showed that the Q-learning agent was able to adapt its decisions to cost constraints, and that including demand information in the 2D model led to slightly more responsive behaviors. While the model operates in a simplified environment, it offers meaningful perspectives on how learning-based systems could support more responsive inventory decisions.