Code for Machine-Learning-Based Discovery of Relaxor Ferroelectric Ceramics

This repository contains the Python code used for machine-learning-based discovery of relaxor ferroelectric materials. The code implements a complete workflow including: • preprocessing of relaxor materials datasets • leakage-safe feature engineering and normalization • fold-wise imputation for processing variables • ensemble neural network training and prediction • uncertainty-aware prediction aggregation • high-throughput compositional screening of perovskite materials