Machine learning based prediction of heavy metals adsorption onto clay
Machine learning based prediction of heavy metals adsorption onto clay
Rasoul Keshmiri Naqab1 Masoud Rabieian2 Reza Arianfar3 Mohsen Taghavijelooudar4
1) Department of Civil and Environmental Engineering Faculty, Babol University of Technology, P.O. Box: 47148-7313, Babol, Iran
2) Department of Civil and Environmental Engineering Faculty, Babol University of Technology, P.O. Box: 47148-7313, Babol, Iran
3) Department of Computer Engineering, Islamic Azad University, Science Research Branch, Tehran, Iran
4) PhD from Department of Civil and Environmental Engineering, Seoul National University, 151-744, Seoul, South Korea
IOI Code
:
AAGF-ABBHZB
Publication :
9th International Conference on Applied Researches in Science & Engineering (carseconf.com/9th)
Abstract :
As a natural adsorbent, clay has been widely used for removing heavy metals. However, reported adsorption capacities of clay minerals vary substantially across studies due to differences in clay characteristics, heavy-metal cation properties, and adsorption conditions. To improve prediction accuracy for heavy-metal adsorption capacity, this study evaluated six machine-learning (ML) regression algorithms to systematically model the adsorption behavior of clay-based materials. Among the tested models, the random forest (RF) algorithm showed the best performance (R² = 0.94; RMSE = 4.47) and was selected as the optimal predictor. Model interpretability analysis further clarified the relative importance of input variables and their influence on adsorption capacity. Feature groups ranked in descending order of importance were adsorption conditions, metal-cation properties, and clay properties. Overall, this work provides deeper insight into heavy-metal adsorption onto clay minerals and offers practical guidance for enhancing adsorption efficiency and supporting remediation of heavy-metal contamination.
Keywords : machine learning adsorption clay heavy metals