Heavy Metal Adsorption Using Low-Cost Adsorbents: A Random Forest Machine Learning Approach For Data Analysis And Performance Prediction
DOI:
https://doi.org/10.64252/b8asez20Keywords:
Adsorption, Heavy metals, Tea waste, Coffee waste, Water treatment, Environmental remediation, Lowcost adsorbents Machine learning, Random Forest, Feature importanceAbstract
Industrial wastewater contamination with heavy metals poses a significant environmental and health challenge in the 21st century. This study investigates the effectiveness of tea and coffee waste as low-cost adsorbents for removing heavy metals (lead, nickel, cadmium, zinc, copper, and iron) from aqueous solutions. The research demonstrates that these agricultural wastes can achieve remarkable removal efficiencies, with lead showing the highest adsorption rate of up to 99.1% under optimal conditions. The study examined various parameters including initial metal concentration (5-30 mg/L), adsorbent dosage (2-3 gm), and contact time (15-60 minutes). To complement the experimental investigation, a Random Forest machine learning model was implemented to analyse the relative importance of operational parameters on metal removal efficiency. The machine learning analysis revealed that contact time emerged as the most influential factor across all metals (importance scores: 0.90-1.82), while initial concentration and adsorbent dose showed varying importance depending on the specific metal. This data-driven approach provided quantitative insights into parameter optimization and validated the experimental findings through predictive modelling. Results indicate that tea and coffee waste represent economical and environmentally sustainable alternatives to conventional treatment methods, with optimal pH range of 4.5-8.0 for maximum metal removal efficiency.