Chemometrics assisted method for classification of mango juice by FTIR spectroscopic data

Abstract

Commercial mango juices are adulterated with heavy use of simple sugars in Bangladesh which poses a serious threat to public health. The present study is aimed to develop chemometrics assisted method for classification of commercial mango juices as adulterated or not with excessive use of glucose, fructose and sucrose with FTIR spectral data. Two statistical techniques, Artificial Neural Network (ANN) and Partial Least Squares-Discriminant Analysis (PLS-DA) have been assessed for their efficiencies in classification in this regard. Before calibration, spectral data were preprocessed with de-noising techniques, Savitzky-Golay (S-G) filtering. Concentration of simple sugars were classified as within or over certain limits. Here spectral values of 64 synthetic mixture solutions are used as training data to develop models and 15 spectral data of real mango juice are used as test data. PLS-DA shows better classification performance over lowercase ANN. From the findings, we develop a method for classification of mango juices adulterated with heavy use of simple sugars (glucose, fructose and sucrose). Therefore, it is a simple and cheap method to classify mango juices as adulterated or safe for consumers, manufacturers and quality regulating authorities.

Bangladesh J. Sci. Ind. Res. 52(2), 73-80, 2017