Adsorption of chromium ions from aqueous solution onto lignin was optimized and modeled under a wide variety of physicochemical factors using machine learning (ML) algorithms. Adsorbent lignin was extracted from jute stick, characterized by different analytical methods, and used for the removal of Cr(VI) ion from aqueous solution. The effects of initial concentration, dosages, pH, duration of adsorption, and temperatures on the adsorption process were studied in batch experiments to determine the optimal parameters. Lignin demonstrated a maximum adsorption capacity of 97.06 mg g -1 at pH 2, and a lignin dose of 50 mg L -1 . Various isotherm and kinetic models were used to interpret the data, and the findings established that the adsorption process is best described by the Langmuir isotherm (R 2 = 0.987), and pseudo-first-order kinetic model (R 2 = 0.978). The experiment was used to train ML algorithms to find out the relative importance of the influencing factors and to identify the most significant parameter governing Cr(VI) adsorption onto lignin. Four ML models, i.e., random forest (RF), extreme gradient boosting (XGBoost), artificial neural networks (ANNs), and k-nearest neighbors (KNNs), were employed for predictive analysis, determining optimal adsorption conditions, and identifying the most influential parameter. Among the four ML models, KNN and XGBoost provided the highest accuracy for optimizing Cr(VI) adsorption onto lignin. The ML results consistently identified adsorbent dosage as the most promising parameter, followed by temperature and initial concentration.
Ahmed et al. (2026) studied this question.