A Comprehensive Analysis of Clustering Propensity of Model Parameters of Voltammetric System

This work reports a comprehensive study of the clustering ability of parameters of the recently developed models of the voltammetric (E-Tongue) system. The E-Tongue setup used in this work has been designed and used earlier for quality analysis of black tea samples. In this work, the parameters of the models are treated as features of the tea samples and they are subjected to principal component analysis (PCA) to assess their clustering tendency. The work is aimed at the identification of the most suitable model that is able to capture the information of the analyte and based on that, the model may be used for future E-Tongue applications. The results suggest that it is the linear equivalent circuit model which came out to be superior than the existing nonlinear and fractional-order model in terms of clustering propensity when PCA was performed with 20 samples of black tea. Therefore, the idea of considering the identified equivalent electric circuit parameters as characteristic features of an analyte is justified in this work.