🤖 AI Summary
This work proposes a robust machine learning workflow to address the limited generalization capability that often undermines reliable prediction of the thermoelectric figure of merit (zT) in Half-Heusler compounds. By employing PCA-based unbiased data splitting to ensure chemical representativeness of the test set, the framework integrates k-best feature selection, Bayesian hyperparameter optimization, and multi-model ensembling. Physical interpretability is enhanced through SHAP and SISSO analyses, shifting the evaluation focus from conventional metrics toward generalization reliability. The approach identifies A-site doping concentration and enthalpy of vaporization as key descriptors for zT and efficiently screens over 660 million candidate structures, yielding multiple stable, high-zT materials that validate the framework’s effectiveness and practical utility.
📝 Abstract
Machine learning (ML) can facilitate efficient thermoelectric (TE) material discovery essential to address the environmental crisis. However, ML models often suffer from poor experimental generalizability despite high metrics. This study presents a robust workflow, applied to the half-Heusler (hH) structural prototype, for figure of merit (zT) prediction, to improve the generalizability of ML models. To resolve challenges in dataset handling and feature filtering, we first introduce a rigorous PCA-based splitting method that ensures training and test sets are unbiased and representative of the full chemical space. We then integrate Bayesian hyperparameter optimization with k-best feature filtering across three architectures-Random Forest, XGBoost, and Neural Networks - while employing SISSO symbolic regression for physical insight and comparison. Using SHAP and SISSO analysis, we identify A-site dopant concentration (xA'), and A-site Heat of Vaporization (HVA) as the primary drivers of zT besides Temperature (T). Finally, a high-throughput screening of approximately 6.6x10^8 potential compositions, filtered by stability constraints, yielded several novel high-zT candidates. Breaking from the traditional focus of improving test RMSE/R^2 values of the models, this work shifts the attention on establishing the test set a true proxy for model generalizability and strengthening the often neglected modules of the existing ML workflows for the data-driven design of next-generation thermoelectric materials.