Regression-Based EGCG Detection in Green Tea Employing MIP Electrodes

In this treatise, multiple tree-based ensemble regression machine learning (ML) approaches were used to estimate the epigallocatechin-3-gallate (EGCG) content (mg/g) in green tea samples. Co-polymerizing ethylene glycol dimethacrylate (EGDMA) with acrylonitrile (AN) yielded the molecularly imprinted polymer EGCG (MIP-EGCG) electrode. On imbibing the MIP-EGCG electrode with real samples, the differential pulse voltammetry (DPV) signatures were recorded to identify performance. The recorded dataset is subjected to decision trees (DTs) and random forest (RF) along with boosting methods namely adaptive boosting (AdaBoost), gradient boost, and extreme gradient boost (XGBoost). Further, hyperparameter adjustments were performed to construct a generalized model. The mean-squared error (mse), mean-squared logarithmic error (MSLE), mean absolute error (MAE), median absolute error (MedAE), and coefficient of determination (
${R}^{{2}}$
) are involved in evaluating the performance of the model. In the context of quantitative prediction of EGCG, the XGBoost algorithm performed well with the value of the coefficient of determination (
${R}^{{2}}$
) being 0.99 and mse value of 0.0001.