Demand Forecast OptimizationusingMachine Learning Algorithms

Sammanfattning

Demand forecast accuracy is a crucial factor when it comes to meeting customer demands and satisfaction while also boosting financial performance. The current logistic and supply chain framework consists of a lot of layers and dependencies, especially after the pandemic, so it is highly important that there should be an accurate prediction. Considering the demand and supply uncertainty is very crucial when it comes to decision-making. The thesis explores the development and implementation of demand forecasting models leveraging four different methods, which include both traditional statistical methods and machine learning methods. The machine learning algorithms used include the Random Forest algorithm, XGBoost algorithm, and an Ensemble algorithm. Out of these methods that were tested out, the Random Forest algorithm went on to have the highest accuracy in prediction for the present study. The model is trained using historical demand data across a time period of two and a half years for a particular product line, from which the demand forecast is generated for the year 2025. The thesis also explores the possible challenges when it comes to the implementation of new-age technologies like machine learning and artificial intelligence in a global organization.