Machine Learning Prediction of Lattice Parameters in Perovskite Oxygen Electrodes for SOECs

Since their discovery, perovskite oxides have been widely studied for applications such as energy storage, fuel cells, and solar cells. Their crystal structure, defined by lattice parameters and atomic positions, governs material stability, critical for solid oxide electrochemical cells. This work aims to develop machine learning models comparing three different tree-based approaches, Decision Tree, Random Forest and XGBoost, evaluating how their different complexity affects performance in predicting the lattice constant of cubic ABO 3 perovskite oxides with cations in two different oxidation states (A 3+ B 3+ and A 2+ B 4+ ). These models have been trained and tested with databases extracted from publications. The evaluation metrics for such models are Mean Absolute Error, Root Mean Squared Error and R 2 score, k-fold cross-validation was implemented for the testing of all models and mean ± standard deviation for all metrics was reported. Also, Multiple Linear Regression was included as a baseline model for performance comparison. The models used ionic radius, electronegativity, and valence of the A and B cations as features and statistical analysis of the target distributions informed model selection, highlighting differences in linear and nonlinear behavior across datasets. XGBoost consistently delivered the best predictive performance between tree models, demonstrating robustness and suitability for future applications.