Development of ions adsorption onto nanoparticles from water/wastewater sources via novel nanocomposite materials: A machine learning-based approach

Abstract

In current decades, adsorption process of prevalent pollutants from water/wastewater source has been of paramount attention. To improve the efficiency of pollutants adsorption, various types of nanocomposite materials have been employed. The use of machine learning-based models is a highly effective and promising approach for analysing data obtained from experimental investigations. This study involves the use of three distinct regression models, namely K-nearest neighbours (KNN), Boosted multilayer perceptron (MLP), and Boosted K-nearest neighbours for the purpose of regression analysis on a limited dataset consisting of two inputs and two outputs. The input features are C0 (Initial concentration) and the type of ion, while the output parameters are Ce (equilibrium concentration) and Qe (adsorption amount). By utilizing these regression models, the study aims to extract useful insights from the available data. By tuning their hyper-parameters, the final models have been carried out, then, evaluated through various metrics. Boosted MLP and Boosted KNN both have R2-score values greater than 0.998. Also, when it comes to MAE (Mean absolute error), Boosted MLP shows 0.0755 for Ce which is more accurate than the other previously implemented methods. As a result, the Boosted MLP model illustrates the same optimized value with the dataset: (Ion = Nickel, C0 = 250, Ce = 206.0). Moreover, it is varied for Qe and equals (Ion = Mercury, C0 = 238.11, Ce = 528.52). The results indicated that machine learning models are promising in adsorption science for prediction and correlation of solute concentration data to optimize the process.