Comparative analysis of edible oils classification using Fourier transform infrared and fluorescence spectroscopy coupled with chemometrics

The classification and identification of various edible oils by different spectroscopic techniques coupled with chemometrics were reported. The practices of mislabeling and adulteration have been commonly practised for economic benefits by manufacturers and vendors. Chemometrics is applied for the comparative study of the classification of eight different edible oils using datasets obtained from fluorescence and Fourier -transform infrared spectroscopy (FTIR). The FTIR and fluorescence spectra datasets of 108 sample readings of eight edible oils were used. The principal component analysis (PCA), and support vector machine (SVM) were implemented to process these data and discriminate edible oils, respectively. PCA showed a clustering trend of all the edible oils in orthogonal space. The accuracy rate of the training and prediction dataset was estimated after SVM analysis. The classification report of the training and prediction dataset showed an accuracy of 100% for fluorescence spectroscopy, whereas, for the FTIR dataset, it was 100% for the training set and 93% for the prediction set. This study showed that fluorescence-SVM emerges as the slightly superior choice for the classification of eight types of edible oils, due to its higher test accuracy.