Chitosan-Based Flocculant Heavy Metal Removal Prediction Using Machine Learning Models.

Area of Science:

• Environmental Science
• Water Treatment Technologies
• Computational Chemistry

Background:

• Heavy metal contamination poses significant environmental and public health risks.
• Effective wastewater treatment requires precise monitoring and remediation strategies.
• Chitosan-based flocculants (CBFs) show promise for heavy metal removal.

Purpose of the Study:

• To evaluate novel machine learning (ML) models for predicting heavy metal (HM) removal efficiency using CBFs.
• To assess the performance of Gradient Boosting Regressor (GBR), Hist Gradient Boosting Regressor (HGBR), Random Forest Regressor (RFR), and Extreme Gradient Boosting Regressor (XGBR).
• To enhance ML model accuracy by incorporating K-means clustering labels.

Main Methods:

• Developed four ML models (GBR, HGBR, RFR, XGBR) using a dataset of 484 flocculation experiments.
• Included K-means clustering labels as an additional feature for improved model learning.
• Tested models on predicting the removal of cadmium (Cd²⁺), copper (Cu²⁺), nickel (Ni²⁺), lead (Pb²⁺), and zinc (Zn²⁺).

Main Results:

• The HGBR model demonstrated superior performance in combined HM removal (R² = 0.94/0.97 for testing/training).
• All models achieved high accuracy for individual metal removal, particularly for nickel (Ni²⁺).
• The GBR model exhibited the lowest error rate for individual metal testing.

Conclusions:

• The HGBR model is a reliable tool for environmental monitoring due to its robust generalization capabilities.
• ML models show significant potential for optimizing HM removal processes in wastewater treatment.
• Future work should focus on integrating these models into real-time monitoring systems and exploring wider environmental applications.