Assessment of M5 model tree for prediction of azithromycin antibiotic removal by multi-wall carbon nanotubes in a fixed-bed column system

Abstract

Abstract In this research, an M5 model tree is employed for the prediction of removal efficiency of azithromycin antibiotics by multi-wall carbon nanotubes (MWCNTs), based on experimental data sets from a laboratory column mode. The effect of total flow time (0–260 min), influent flow rates (0.5, 1, and 1.5 mL min−1), bed depths (2, 4, and 6 cm), initial azithromycin concentrations (25, 50, and 100 mg L−1), and pHs (2, 4, 6, 8, and 10) was considered in the adsorption process. Based on the obtained structures, three linear equations (LM, LM2, and LM3) were developed. The root mean square error (RMSE) of 9.89% and determination coefficient (R2) of 0.946 were determined for predicting azithromycin removal by the M5 model tree. The results indicated that contact time was more important in the adsorption process, relative to other operating conditions. This research showed that the M5 model tree could be an accurate and faster alternative to the available mathematical models to estimate removal rates of pollutants. The results obtained from the FTIR technique confirmed that the O–H groups on the MWCNTs surface have an important role in azithromycin adsorption.