This paper presents an eXtreme Gradient Boosting (XGBoost) machine-learning (ML) model for designing Fe-based metallic glasses (MGs) with balanced saturation flux density (Bs) and thermal stability. The model, using intrinsic elemental properties (atomic size and electronegativity), predicts Bs and Tx (onset crystallization temperature) with 93.0% and 94.3% accuracy, respectively. Key features dictating Bs and Tx are identified, revealing physical origins of high Bs and thermal stability. The model successfully guided the development of Fe-based MGs with high Tx (>800 K) and high Bs (>1.4 T), demonstrating the interpretability and feasibility of the XGBoost approach for designing high-performance glassy materials.
