A modified sequential wavenumber selection-discriminant analysis with Bayesian optimization strategy for detection and identification of chia seed oil adulteration using Raman spectroscopy

Saman Abdanan Mehdizadeh,Mohammad Noshad,Mohammad Hojjati

DOI: https://doi.org/10.1016/j.talanta.2024.126439

IF: 6.1

2024-06-20

Talanta

Abstract:The detection of oil fraud can be accomplished through the use of Raman spectroscopy, which is a potent analytical technique for identifying the adulteration of edible oils with inferior or less expensive oils. However, appropriate data reduction and classification methods are required to achieve high accuracy and reliability in the analysis of Raman spectra. In this study, data reduction algorithms such as principal component analysis (PCA) and modified sequential wavenumber selection (MSWS) were applied, along with discriminant analysis (DA) as a classifier for detecting oil fraud. The parameters of DA, such as the discriminant type, the amount of regularization, and the linear coefficient threshold, were optimized using Bayesian optimization. The methods were tested on a dataset of chia oil mixed with 5–40 % sunflower oil, which is a common form of fraud in the market. The results showed that MSWS-DA achieved 100 % classification accuracy, while PCA-DA achieved 91.3 % accuracy. Therefore, it was demonstrated that Raman spectroscopy combined with MSWS-DA and Bayesian optimization can effectively detect oil fraud with high accuracy and robustness.

chemistry, analytical