ЭМПИРИЧЕСКОЕ СРАВНИТЕЛЬНОЕ ИССЛЕДОВАНИЕ ГРАДИЕНТНОГО БУСТИНГА ДЛЯ ПРОГНОЗИРОВАНИЯ ТЕРМИЧЕСКОГО РАЗЛОЖЕНИЯ В ПОЛИМЕРНЫХ КОМПОЗИТАХ С МИНЕРАЛЬНЫМИ НАПОЛНИТЕЛЯМИ

В данном исследовании представлен математический и эмпирический анализ алгоритмов градиентного бустинга, применяемых к данным термогравиметрического анализа (ТГА) композитов на основе винилэфирных и эпоксидных смол с минеральными наполнителями. Алгоритм градиентного бустинга пошагово представлен на численном примере ТГА. Далее формально проанализированы алгоритмы XGBoost и CatBoost. Сравнение 12 алгоритмов выполнено на основе 478 данных из рецензируемой литературы с использованием пятикратно повторенной 5-блочной кросс-валидации. Методы Extra Trees и CatBoost показали статистически эквивалентные наилучшие результаты (среднее R² = 0.7430 ± 0.0879 и 0.7313 ± 0.0793 соответственно; Nadeau & Bengio corrected t = 0.483, p = 0.634), превзойдя все линейные базовые модели, SVR и MLP. Для практического применения рекомендуется CatBoost: его упорядоченный бустинг обеспечивает несмещенные оценки градиентов на малых выборках (n < 1000), а SHAP-анализ структуры градиентных деревьев выявляет физически интерпретируемые атрибуции признаков, согласующиеся с известными механизмами ТГА. Значения SHAP показали, что пиковая температура TGA (T_max) и тип матрицы являются основными определяющими факторами. Предложенный пайплайн предварительной обработки, включающий преобразование phr в wt%, label encoding и медианную импутацию, обеспечивает воспроизводимую стратегию для разнородных литературных TGA-данных.

1. Karuppusamy, M., Thirumalaisamy, R., Palanisamy, S., Nagamalai, S., El Sayed Massoud, E., and Ayrilmis, N. A review of machine learning applications in polymer composites: advancements, challenges, and future prospects. Journal of Materials Chemistry A, 13, 16290–16308 (2025). https://doi.org/10.1039/D5TA00982K

2. Martin, T.B., and Audus, D.J. Emerging trends in machine learning: a polymer perspective. ACS Polymers Au, 3, 239–258 (2023). https://doi.org/10.1021/acspolymersau.2c00053

3. Pai, S.M., Shah, K.A., Sunder, S., Albuquerque, R.Q., Brütting, C., and Ruckdäschel, H. Machine learning applied to the design and optimization of polymeric materials: a review. Next Materials, 7, 100449 (2025). https://doi.org/10.1016/j.nxmate.2024.100449

4. Jafari, P. et al. Machine learning for expediting next-generation of fire-retardant polymer composites. Composites Communications, 45, 101806 (2024). https://doi.org/10.1016/j.coco.2023.101806

5. Xiao, J., Hobson, J., Ghosh, A., Haranczyk, M., and Wang, D.-Y. Flame retardant properties of metal hydroxide-based polymer composites: a machine learning approach. Composites Communications, 40, 101593 (2023). https://doi.org/10.1016/j.coco.2023.101593

6. Berladir, K., Antosz, K., Ivanov, V., and Mitaľová, Z. Machine learning-driven prediction of composite materials properties based on experimental testing data. Polymers, 17, 694 (2025). https://doi.org/10.3390/polym17050694

7. Friedman, J.H. Greedy function approximation: a gradient boosting machine. Annals of Statistics, 29, 1189–1232 (2001). https://doi.org/10.1214/aos/1013203451

8. Chen, T., and Guestrin, C. XGBoost: a scalable tree boosting system. In: Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, ACM, San Francisco, pp. 785–794 (2016). https://doi.org/10.1145/2939672.2939785

9. Prokhorenkova, L., Gusev, G., Vorobev, A., Dorogush, A.V., and Gulin, A. CatBoost: unbiased boosting with categorical features. arXiv:1706.09516 (2019). https://doi.org/10.48550/arXiv.1706.09516

10. Malashin, I., Tynchenko, V., Gantimurov, A., Nelyub, V., and Borodulin, A. Boosting-based machine learning applications in polymer science: a review. Polymers, 17, 499 (2025). https://doi.org/10.3390/polym17040499

11. Ge, W., De Silva, R., Fan, Y., Sisson, S.A., and Stenzel, M.H. Machine learning in polymer research. Advanced Materials, 37, 2413695 (2025). https://doi.org/10.1002/adma.202413695

12. Han, J. et al. Toward accurate machine learning-driven prediction of polymeric composites properties based on experimental data. Materials Genome Engineering Advances, 3, e70027 (2025). https://doi.org/10.1002/mgea.70027

13. Xu, P., Ji, X., Li, M., and Lu, W. Small data machine learning in materials science. npj Computational Materials, 9, 42 (2023). https://doi.org/10.1038/s41524-023-01000-z

14. Zhong, X., Gallagher, B., Liu, S., Kailkhura, B., Hiszpanski, A., and Han, T.Y.-J. Explainable machine learning in materials science. npj Computational Materials, 8, 204 (2022). https://doi.org/10.1038/s41524-022-00884-7

15. Breiman, L., Friedman, J.H., Olshen, R.A., and Stone, C.J. Classification and Regression Trees. Routledge, New York (1984). https://doi.org/10.1201/9781315139470

16. Florek, P., and Zagdański, A. Benchmarking state-of-the-art gradient boosting algorithms for classification. arXiv:2305.17094 (2023). https://doi.org/10.48550/arXiv.2305.17094

17. Boldini, D., Grisoni, F., Kuhn, D., Friedrich, L., and Sieber, S.A. Practical guidelines for the use of gradient boosting for molecular property prediction. Journal of Cheminformatics, 15, 73 (2023). https://doi.org/10.1186/s13321-023-00743-7

18. Nadeau, C., and Bengio, Y. Inference for the generalization error. Machine Learning, 52, 239–281 (2003). https://doi.org/10.1023/A:1024068626366

19. Bentéjac, C., Csörgő, A., and Martínez-Muñoz, G. A comparative analysis of gradient boosting algorithms. Artificial Intelligence Review, 54, 1937–1967 (2021). https://doi.org/10.1007/s10462-020-09896-5

20. Cakiroglu, M., Bekdaş, G., and Geem, Z.W. Fatigue predictive modeling of composite materials for wind turbine blades using explainable gradient boosting models. Coatings, 15, 325 (2025). https://doi.org/10.3390/coatings15030325

21. Mustapha, I.B., Abdulkareem, M., Jassam, T.M., AlAteah, A.H., Al-Sodani, K.A., Al-Tholaia, M.M.H., Nabus, H., Alih, S.C., Abdulkareem, Z., and Ganiyu, A. Comparative analysis of gradient-boosting ensembles for estimation of compressive strength of quaternary blend concrete. International Journal of Concrete Structures and Materials, 18, 18 (2024). https://doi.org/10.1186/s40069-023-00653-w

22. Hancock, J.T., and Khoshgoftaar, T.M. CatBoost for big data: an interdisciplinary review. Journal of Big Data, 7, 94 (2020). https://doi.org/10.1186/s40537-020-00369-8

23. Pemila, M., Pongiannan, R.K., Narayanamoorthi, R., AboRas, K.M., and Youssef, A. Application of an ensemble CatBoost model over complex dataset for vehicle classification. PLOS ONE, 19, e0304619 (2024). https://doi.org/10.1371/journal.pone.0304619

24. Sheridan, R.P., Wang, W.M., Liaw, A., Ma, J., and Gifford, E.M. Extreme gradient boosting as a method for quantitative structure–activity relationships. Journal of Chemical Information and Modeling, 56, 2353–2360 (2016). https://doi.org/10.1021/acs.jcim.6b00591

25. Shwartz-Ziv, R., and Armon, A. Tabular data: deep learning is not all you need. Information Fusion, 81, 84–90 (2022). https://doi.org/10.1016/j.inffus.2021.11.011

26. McElfresh, D. et al. When do neural nets outperform boosted trees on tabular data? arXiv:2305.02997 (2023). https://doi.org/10.48550/arXiv.2305.02997

27. Malashin, I.P., Tynchenko, V.S., Nelyub, V.A., Borodulin, A.S., and Gantimurov, A.P. Estimation and prediction of the polymers' physical characteristics using the machine learning models. Polymers, 16, 115 (2024). https://doi.org/10.3390/polym16010115

28. Kucukarslan, O. Machine learning-based heat deflection temperature prediction and effect analysis in polypropylene composites using CatBoost and Shapley Additive Explanations. Engineering Applications of Artificial Intelligence, 120, 106873 (2023). https://doi.org/10.1016/j.engappai.2023.106873

29. Phoeuk, M. et al. Accuracy prediction of compressive strength of concrete incorporating recycled aggregate using ensemble learning algorithms: multinational dataset. Advances in Civil Engineering, 2023, 5076429 (2023). https://doi.org/10.1155/2023/5076429

30. Uddin, M.N., Shanmugasundaram, N., and Praveenkumar, S. Prediction of compressive strength and tensile strain of engineered cementitious composite using machine learning. International Journal of Mechanics and Materials Design, 20, 671–716 (2024). https://doi.org/10.1007/s10999-023-09695-0

31. Khodadadi, N., Roghani, H., De Caso, F., El-kenawy, E.-S.M., Yesha, Y., and Nanni, A. Data-driven PSO-CatBoost machine learning model to predict the compressive strength of CFRP-confined circular concrete specimens. Thin-Walled Structures, 198, 111763 (2024). https://doi.org/10.1016/j.tws.2024.111763

32. Aydın, Y., Cakiroglu, C., Bekdaş, G., and Geem, Z.W. Explainable ensemble learning and multilayer perceptron modeling for compressive strength prediction of ultra-high-performance concrete. Biomimetics, 9, 544 (2024). https://doi.org/10.3390/biomimetics9090544

33. Lundberg, S.M., and Lee, S.-I. A unified approach to interpreting model predictions. Advances in Neural Information Processing Systems, 30, 4766–4777 (2017). https://doi.org/10.48550/arXiv.1705.07874

34. Tsioptsias, C. et al. Simulation and experimental study of the isothermal thermogravimetric analysis and the apparent alterations of the thermal stability of composite polymers. Polymers, 16, 1454 (2024). https://doi.org/10.3390/polym16111454

35. Gurnani, R., Kuenneth, C., Toland, A., and Ramprasad, R. Polymer informatics at scale with multitask graph neural networks. Chemistry of Materials, 35, 1560–1567 (2023). https://doi.org/10.1021/acs.chemmater.2c02991

36. Frie, C., Schork, N., Reiff, A., and Melz, T. Exploration of materials fatigue influence factors using interpretable machine learning. Fatigue and Fracture of Engineering Materials and Structures, 47, 2752–2773 (2024). https://doi.org/10.1111/ffe.14315

37. Elaskalany, M. et al. Stochastic multiscale modeling of electrical conductivity of carbon nanotube polymer nanocomposites: an interpretable machine learning approach. Advanced Engineering Materials, 26, 2401233 (2024). https://doi.org/10.1002/adem.202401233

38. Mykytyshyn, A. et al. Machine learning approaches for classification of composite materials. Modelling, 6, 118 (2025). https://doi.org/10.3390/modelling6040118

39. Wan, Z., Chen, S., Feng, X. & Sun, Z.-Y. From processing to properties: enhancing machine learning models with microstructural information in polymer nanocomposites. Composites Communications, 51, 102072 (2024). https://doi.org/10.1016/j.coco.2024.102072

40. Champa-Bujaico, E., Díez-Pascual, A.M., Redondo, A.L., and Garcia-Diaz, P. Interpretable machine learning framework to predict the glass transition temperature of polymers. Polymers, 16, 1049 (2024). https://doi.org/10.3390/polym16081049

41. Liu, B., Vu-Bac, N., Zhuang, X., Fu, X., and Rabczuk, T. Stochastic integrated machine learning based multiscale approach for the prediction of the thermal conductivity in carbon nanotube reinforced polymeric composites. Composites Science and Technology, 224, 109425 (2022). https://doi.org/10.1016/j compscitech.2022.109425