Akademik Veri Yönetim Sistemi
7th International Conference on Intelligent and Fuzzy Systems, INFUS 2025, İstanbul, Türkiye, 29 - 31 Temmuz 2025, cilt.1529 LNNS, ss.114-122, (Tam Metin Bildiri)
In this study, different machine learning techniques, random forest, gradient boosting, and extreme gradient boosting (XGBoost), are analyzed on the dataset derived from the numerical results of a mixed convection flow problem in a lid-driven cavity in the presence of induced magnetic field in terms of magnetic potential. The problem is numerically solved by radial basis function (RBF) method, and the data is collected from numerical results in different dimensionless problem parameters, Reynolds number (in [10, 1000]), Grashof number (in [100, 100000]), Hartmann number (in [1, 50]) and magnetic Reynolds number (in [1, 25]). The leading problem parameter, average Nusselt number along the heated wall is modeled by the data with these ensemble learning algorithms which utilize decision trees as base learners. Even though the gradient boosting outperforms the random forest in terms of accuracy, the random forest outperforms the gradient boosting in terms of computational efficiency. At that juncture, the XGBoost is selected for the optimum practice as it performs better accuracy than the gradient boosting, and outperforms the random forest in terms of computational efficiency for large datasets. Further, feature importance analyzed in each modeling results indicate that the average Nusselt number is affected by Reynolds number primarily whereas the magnetic Reynolds number produces the minimum, even negligible effect.