Data-driven approaches to wastewater treatment: machine learning for heavy metals adsorption
Year: 2026
Type: article
Abstract: The increasing presence of heavy metal pollutants in the environment poses significant challenges that call for effective removal solutions.However, data collected from various experimental studies often involve different adsorbents and operating conditions, introducing heterogeneity that can reduce model accuracy and complicate result comparisons.This research aims to address these challenges by applying machine learning algorithms-including random forest regression, gradient boosting regression, class boosting regression, and artificial neural networks-to predict the removal rates of heavy metals such as iron, lead, cadmium, copper, nickel, and chromium.The results demonstrate that the gradient boosting regression model outperforms other methods, achieving a high correlation coefficient of 0.957 and a low root mean squared error of 0.0093 for predicting the removal rate.To ensure the model's reliability and robustness, thorough evaluation techniques including feature importance analysis, correlation matrix examination, and k-fold crossvalidation were employed.These assessments confirmed that initial concentration, pollutant type, and pH are the most influential factors impacting removal efficiency, while the model exhibits strong generalizability and stability across different data subsets.Overall, this study highlights the effectiveness of integrating advanced machine learning approaches with rigorous validation to enhance the prediction and optimization of heavy metal removal processes
Source: PPOR
Language: English
Subfield: Water Science and Technology
Field: Environmental Science
Domain: Physical Sciences
Open Access status: bronze
OpenAlex ID: W7127044325