Advanced removal of butylparaben from aqueous solutions using magnetic molybdenum disulfide nanocomposite modified with chitosan/beta-cyclodextrin and parametric evaluation through sequential multi-objective machine learning algorithms

• This study successfully removed butylparaben from water using a beta-cyclodextrin and chitosan-modified magnetic molybdenum disulfide nanocomposite, achieving optimized performance through machine learning. • Polynomial Regression (PR) outperformed Gradient Boosting Regression (GBR) by exhibiting lower error values, with MAE = 3.9, MSE = 26.8, and RMSE = 5.18, ensuring better predictive accuracy. • The L-BFGS optimization method identified optimal conditions as pH = 6.64, initial butylparaben concentration = 1.00 mg/L, contact time = 60 min, and adsorbent dosage = 0.8 g/L, enhancing removal efficiency. • This study highlights the synergistic effect between advanced nanocomposites and machine learning algorithms, offering an effective and scalable solution for sustainable water treatment and industrial process optimization. The research examines the efficient butylparaben (BUP) elimination process from aqueous media through beta-cyclodextrin/chitosan-modified magnetic molybdenum disulfide nanocomposite (BCCMM) utilization. A multi-algorithmic approach involving Gradient Boosting Regression (GBR), Polynomial Regression (PR) together with Limited-memory Broyden-Fletcher-Goldfarb-Shanno (L-BFGS) was used to enhance removal condition optimization. The scientific evaluation studied the factors pH (X1) and initial BUP concentration (X2) contact time (X3) and adsorbent dosage (X4). The model performance was assessed through Mean Absolute Error (MAE), Mean Squared Error (MSE), Mean Absolute Percentage Error (MAPE), Root Mean Squared Error (RMSE) and the statistical metrics R², adjusted R², Explained Variance Score (EVS). The GBR model exhibited higher error values on the test set, particularly in MSE, indicating possible overfitting. Polynomial Regression outperformed the training data prediction with accuracy metrics MAE = 3.9, MSE = 26.8, MAPE= 7.64 and RMSE = 5.18. The predictive stability of PR emerges through these different dataset applications. The L-BFGS algorithm established the optimal control factors as pH = 6.64 and initial concentration = 1.00 mg/L and contact time = 60 min and adsorbent dosage = 0.8 g/L which dramatically improved the removal efficiency due to the collaborative properties of the nanocomposite. This research fully evaluates machine learning approaches that optimize complicated environmental remediation operations while advanced nanomaterials used with data-driven optimization provide a strong adaptable method to remove organic water pollutants thus supporting sustainable treatment development. Butylparaben extraction performance reaches its maximum level while resource utilization improves throughout the entire extraction technique under the established optimal conditions. Such improvements play a vital position in developing large-scale sustainable water remedy methods. Nanomaterials technology combined with data optimization techniques allows the method to create adaptable solutions for organic pollution cleaning at real water pollution locations.