<![CDATA[Metal–Organic Frameworks (MOF) are porous materials with broad potential in energy, catalysis, and gas storage applications; however, their thermal stability remains a critical challenge. This study aims to develop a predictive model for the thermal stability of Zn-MOF based on the Quantitative Structure–Property Relationship (QSPR) using a stacking ensemble regression approach. The model was constructed with six base algorithms—Gaussian Process, KNN Regression, Ridge Regression, Linear Regression, Lasso Regression, and Support Vector Regression (SVR)—which were subsequently combined through TheilSen Regression as a meta-learner. The evaluation results demonstrate excellent predictive performance, with an MAE of 0.0022, an RMSE of 0.0023, and an R² of 0.9994 on the test set. Validation against experimental data also revealed only a tiny prediction error, confirming the model’s ability to accurately represent the complex relationships between structural descriptors and thermal stability. Compared to single regression models, the stacking approach proved to be more stable, robust against outliers, and adaptive to heterogeneous data distributions. These findings highlight the potential of stacking ensembles as an efficient data-driven material design strategy, enabling faster identification and design of thermally stable MOF while reducing reliance on costly and time-consuming laboratory experiments.]]>
Copyrights © 2025
