Accelerated design and discovery of perovskites with high conductivity for energy applications through machine learning

Abstract We use machine learning tools for the design and discovery of ABO 3 -type perovskite oxides various energy applications, using over 7000 data points from literature. demonstrate a robust framework efficient accurate prediction total conductivity perovskites their classification based on type charge carrier at different conditions temperature environment. After evaluating set >100 features, we identify average ionic radius, minimum electronegativity, atomic mass, formation all B-site, B-site dopant ions as crucial relevant predictors determining carriers. The models are validated by predicting compounds absent in training set. screen 1793 undoped 95,832 A-site doped to report with high conductivities, which can be used depending