Data-centric approach to improve machine learning models for inorganic materials

Pandey et al. (2021) demonstrate the importance of diversifying training data to make balanced predictions of thermodynamic properties for inorganic crystals.