Rapid Classification of Coffee Origin Using Near-Infrared (Nir) Spectroscopy and (Non) Linear Machine Learning

Over the past decade, there has been overwhelming interest in rapid and routine origin tracing and authentication methods, such as near-infrared spectroscopy (NIR). In a systematic and comprehensive approach, this study coupled NIR with advanced machine learning models to explore the origin classification of coffee at various scales (continental to regional level). Specialty green coffee beans were sourced from three continents, eight countries, and 22 regions. The dispersive bulk NIR was used for spectral registration in the reflectance mode, and the obtained spectra were preprocessed with extended multiplicative scatter correction and mean centering. The classical chemometrics models (such as PCA, PLS-DA) were first employed. The linear PLS-DA adequately predicted origin at the country level and showed promise at the regional level. Nevertheless, PLS-DA model poorly predicted origin at the continental and the Central American regional levels. Non-linear machine learning models improved predictions with the best accuracy found using random forest with accuracies up to 0.97 to 0.99 for the continental and the Central American models, respectively. Discriminating wavelength regions and constituents were identified at each level. The present work demonstrated the potential of NIR and machine learning data analysis for rapid origin classification of coffee from the continental to the regional level.