Fig. 2. Performance comparison of models across metrics and Area Under the Curve (AUC) curves.

a The hist plot of different models in terms of seven metrics. Source data are provided as a Source Data file. b AUC curves, and (c) zoomed AUC curves of different models. The comparison models include Electron Configuration models with Stacked Generalization (ECSG), Roost⁴⁸, CrabNet⁶⁹, Random Forest (RF), AdaBoost (Ada), Magpie³⁰, Meredig³¹ ElemNet²⁹, Atom Table Convolutional Neural Networks (ATCNN)³⁷ and Electron Configuration Convolutional Neural Network (ECCNN).