Thermal Conductivity and Debye Temperature of Transition Metals Prediction Using Machine Learning Support Vector Machine Models

The lattice thermal conductivity and Debye temperature of transition metals are fundamental properties governing their applications in thermal management, high-temperature alloys, and energy-related technologies, while also providing insight into lattice dynamics and heat-transport behavior. However, accurate prediction of these properties remains challenging due to anharmonic lattice effects, complex structure-property relationships, and limited experimental data availability. In this study, machine learning approaches were employed to predict the lattice thermal conductivity and Debye temperature of transition metals using a dataset containing nearly 1000 records with relevant physical and chemical descriptors. Several Support Vector Machine (SVM)-based regression models, including Linear SVM (LSVM), Quadratic SVM (QSVM), Cubic SVM (CSVM), Fine Gaussian SVM (FGSVM), Medium Gaussian SVM (MGSVM), and Coarse Gaussian SVM (CGSVM), were systematically investigated. Among the evaluated models, QSVM and CSVM demonstrated superior performance for Debye temperature prediction, achieving prediction accuracies of 96.80% and 99.01% on validation and test datasets, respectively. For lattice thermal conductivity prediction, CSVM outperformed all competing models, reaching accuracies of 94.67% and 98.02% on validation and test sets, respectively. Furthermore, the RReliefF algorithm was employed to identify the most influential descriptors governing the target properties. Partial dependence analysis and correlation studies revealed strong coupling between Debye temperature and lattice thermal conductivity and provided physically interpretable insights into feature influence. The results demonstrate the capability of machine learning for accurate prediction and accelerated design of transition metals with tailored thermophysical properties.