Decision Support Tool for Dynamic Inventory Management using Machine Learning, Time Series and Combinatorial Optimization

Nowadays, the imbalance between supply and demand increases the volatility of many raw materials for both production and consumption. In this regard, stock management inventory would be necessary in order to limit losses and increase margins. It this perspective, we propose through this article, the development of a model combining time series, machine learning algorithm and combinatorial optimization in order to identify the opportunities to buy stock at a lower cost and to sale a portion of the unused stock to generate additional profits for the organization. In this model, we use machine learning models and time series to predict stock prices and demand forecast of stock to ensure the production and then integrate them into a combinatorial optimization algorithm to help making the best decision (Buy, Sell or Hold), and the exact quantities to buy or to sell in order to maximize earnings without risks.