Explore our research and dive into our codebase. Feel free to browse around, and don’t hesitate to reach out if you have any questions or need more information!
Due to shipping delays and stockouts, inventory management inefficiency leads to profit loss, missed sales opportunities, and reduced customer satisfaction. Having the right amount of inventory is critical to improving business operations. DataCo Global, a multinational company distributing clothes, sports equipment, and electrical supplies, provides an example of how data can solve stockout issues.
The causes of stockouts and poor inventory management are related to poor-performing suppliers, inefficient demand forecasting, fewer workers, natural disasters, and fluctuating customer demands. These factors cause significant challenges in maintaining optimal inventory levels.
Organizations like Amazon have addressed these problems by successfully implementing machine learning tools to optimize inventory levels and reduce stock. However, most companies and data scientists have approached solving these problems mathematically, not levering deep reinforcement learning. Deep reinforcement learning models thrive in complex environments, are adaptive decision-makers, and can learn from real-time data, which makes these models suitable for solving complex problems like inventory management.
Our capstone project focuses on optimizing inventory efficiency. Our goal is to mitigate stockouts, and we will do this by using the data of the company DataCo. We will also use advanced statistical concepts like Deep Reinforcement Learning to tackle this problem.
Q1) How can improved forecasting techniques reduce stockouts in inventory management?
Q2) What are the key factors that influence benefit (profit)?
Q3) How can we predict if a product from a specific department will be delivered on time?
Through the models we developed, we identified effective strategies to maximize rewards, primarily revenue, while minimizing various costs. Additionally, we uncovered key variables that significantly influenced profitability, confirming that the company is generating profits and is not in financial distress. We were also able to predict whether a product would be shipped on time or face delays. However, we found that the dataset used in this study was not ideal and lacked critical information needed to enhance the model's effectiveness and enable its practical implementation in real-world scenarios. In conclusion, this model serves as a foundational benchmark for developing future models utilizing deep reinforcement learning.