Predictive Capability of Dye Removal from Wastewater Using Biochar by a Rough Set Machine Learning Model

Abstract

Dye removal from wastewater treatment plants has gained attention in the waste management sector, necessitating advanced prediction techniques for effective planning and execution. While numerous machine learning studies have explored dye removal using biochar, a lack of general rules for various wastewater sources remains. This study employs rough set machine learning (RSML) to predict dye removal based on decision attributes, generating IF-THEN rules to classify conditional attributes. Key attributes identified include solution pH, temperature, and the initial concentration ratio of biochar to dye, which are critical for accurate predictions of the dye removal efficiency. The model produced 45, 23, and 39 rules for methylene blue, crystal violet, and Congo red, respectively, with 14, 11, and 15 approximate rules. The RSML achieved more than 80% accuracy for all three dyes, outperforming the existing classifiers. These findings have significant implications for establishing scientific rules in future dye removal research using biochar adsorption, enhancing the effectiveness of wastewater treatment processes.