Development of a Generalized Statistical Model for Hexavalent Chromium Removal Using Electrocoagulation Through SVR Regression Analysis

Abstract

Hexavalent chromium Cr(VI) is a highly toxic pollutant that poses a significant threat to human health and the environment. Electrocoagulation is a promising technology for the removal of Cr(VI) from wastewater. This work reviews and evaluates statistical models developed in different studies published between 2015 and 2021 on the removal of Cr(VI) using electrocoagulation. The analysis showed that none of the models was found to be conclusive, and that they all suffer from issues such as overfitting and the inability to generalize beyond the experiment domain. These models were also highly dependent on the selection of input parameters, model selection criteria, and experimental design. An attempt to solve this problem was to utilize Machine Learning (ML) techniques to develop a more robust model that can provide generalized and accurate predictions on a broader domain. The model was developed using Support Vector Machines Regression analysis (SVR). Data compiled from previously published works were used to train and test the model using a 50:50 split ratio. The model was able to make more generalized predictions but lacked accuracy. As with all ML models, this model requires a higher volume of high-quality data to improve its accuracy. The study concluded that there is still a need for more robust statistical models that can effectively capture the complexity of the electrocoagulation process and generalize well beyond the experiment domain.