Comparison of NIR spectroscopy devices (Portable and Benchtop) and a low-cost E-nose for classification of black tea: A machine learning approach using PLSDA, LDA, and PCA

Black tea adulteration is a widespread practice, involving the mixing or substitution of high-quality tea leaves with lower quality samples. The authentication of black tea regarding its composition and origin is crucial to ensure quality and maintain consumer trust. However, conventional methods (e.g., High-Performance Liquid Chromatography HPLC) for tea analysis are expensive, demand chemicals, and generate waste. This study evaluates two NIR instruments (portable (PNIR) and benchtop (BNIR)) and low-cost electronic nose (LC-e-nose) for classification of tea leaves from three different origins: Brazil (BR), United States (US), and India (IND). Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA), and Partial Least Squares Discriminant Analysis (PLS-DA) were applied for data treatment, and the results demonstrate that the LC-e-nose exhibits F1 scores above 99% for all models, comparable to NIR systems (F1 > 99%). These findings highlight its potential as a cost-effective and reliable alternative to NIR spectroscopy for investigating tea quality and origin. By comparing these advanced techniques, this study provides an evaluation of black tea quality and authenticity at a significantly reduced cost. • LC-e-nose achieved >99% F1 scores, rivaling traditional NIR systems. • Portable and benchtop NIR instruments effectively classified tea by origin. • Combining low-cost NIR and LC-e-nose enhances tea quality and authenticity. • Our LC-e-nose can handle unbalanced data effectively