Ensemble methods and their interpretability in demand forecasting

ai generated, neural, network-8540914.jpg
Publication notice: 14th Joint Conference on Mathematics and Computer Science, Cluj-Napoca
Conference: 14th Joint Conference on Mathematics and Computer Science, Cluj-Napoca

We are thrilled to announce the recent publication of research by our partner University Babeș-Bolyai done for the Optimal DataSeer project, at the 14th Joint Conference on Mathematics and Computer Science. The research focuses on using ensemble models for demand prediction and applying various interpretability techniques to gain insights into these models. Different boosting ensemble methods like Adaboost, XGBoost, and bagged decision trees such as Random Forest were evaluated for solving regression problems in demand forecasting. Techniques like SHAP tree explainer, permutation feature importance, and Mimic explainer were employed to provide feature importance values, offering both global and local explanations.

Scroll to Top