Heavy metal adsorption efficiency prediction using biochar properties: a comparative analysis for ensemble machine learning models DOI Creative Commons
Zaher Mundher Yaseen‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬,

Farah Loui Alhalimi

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: April 18, 2025

The contamination of water and soils with heavy metals poses a significant environmental threat, making the development effective removal strategies global priority. Hence, determination can play an essential role in monitoring assessment. In current research, ensemble machine learning (ML) models (i.e., Random Forest Regressor (RFR), Adaptive Boosting (Adaboost), Gradient (GB), HistGradientBoosting, Extreme (XGBoost), Light Gradient-Boosting Machine (LightGBM)) were applied attempt to predict adsorption efficiency several Pb, Cd, Ni, Cu, Zn) according different factors including temperature, pH, biochar characteristics. Data collected from open-source literature review 353 samples. At first stage, data processing was performed outliers' scaling for better modeling applicability; whereas, second stage predictive conducted. results showed that XGBoost model attained superior accuracy comparison other by achieving highest coefficient (R2 = 0.92). research extended investigate feature importance analysis which indicated initial concentration ratio pH most influential toward followed Pyrolysis while features like physical properties as surface area pore structure had minimal effect on efficiency. These findings highlighted using ML guiding solutions it provides efficient prediction ease selection application.

Language: Английский

Heavy metal adsorption efficiency prediction using biochar properties: a comparative analysis for ensemble machine learning models DOI Creative Commons
Zaher Mundher Yaseen‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬,

Farah Loui Alhalimi

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: April 18, 2025

The contamination of water and soils with heavy metals poses a significant environmental threat, making the development effective removal strategies global priority. Hence, determination can play an essential role in monitoring assessment. In current research, ensemble machine learning (ML) models (i.e., Random Forest Regressor (RFR), Adaptive Boosting (Adaboost), Gradient (GB), HistGradientBoosting, Extreme (XGBoost), Light Gradient-Boosting Machine (LightGBM)) were applied attempt to predict adsorption efficiency several Pb, Cd, Ni, Cu, Zn) according different factors including temperature, pH, biochar characteristics. Data collected from open-source literature review 353 samples. At first stage, data processing was performed outliers' scaling for better modeling applicability; whereas, second stage predictive conducted. results showed that XGBoost model attained superior accuracy comparison other by achieving highest coefficient (R2 = 0.92). research extended investigate feature importance analysis which indicated initial concentration ratio pH most influential toward followed Pyrolysis while features like physical properties as surface area pore structure had minimal effect on efficiency. These findings highlighted using ML guiding solutions it provides efficient prediction ease selection application.

Language: Английский

Citations

1