A systematic and critical review on development of machine learning based-ensemble models for prediction of adsorption process efficiency

The development of machine learning-based ensemble models for the prediction of complex processes with non-linear nature (such as adsorption) has been remarkably advanced over recent years. As a result, having an informative vision of these models' progression, appears to be critical for better understanding and using them in applications such as adsorption modeling. This paper systematically and critically reviews 38 articles in the field of application of ensemble models for the prediction of adsorption process efficiency for pollutants' removal from aquatic solutions. Two aspects, including the adsorption process and ensemble models’ characteristics, are discussed in details. The type of adsorbate and adsorbent, as well as the system operation mode, are explored from the first point of view. The type of ensemble technique, software, input and output variables, dataset size and partitioning method, and performance metrics are all investigated in the ensemble model section. Based on discussed aspects and outcomes acquired from reviewed papers, some future research perspectives, including choosing model input variables from adsorbate properties, adsorbent characteristics, and adsorption condition parameters to increase the reliability of model predictions and also increasing dataset size to augment the accuracy of the ensemble models, are recommended for promoting next investigations.