Research on the identification of turmeric from different origins based on portable near-infrared spectroscopy combined with chemometrics

Turmeric is recognized worldwide both as a culinary spice and as an herbal remedy. Studies have reported significant variation in its chemical composition depending on its geographical source. Some suppliers falsely claim premium origins to pass off lower-quality products, driven by profit motives. To address this, there is a growing need for a rapid, practical method to verify the origin of turmeric. In this study, portable near-infrared spectroscopy (NIRS), combined with various chemometric methods, was used to identify turmeric from 9 different regions. The local outlier factor (LOF) method outperformed for detecting and removing outliers. Among preprocessing techniques, the first derivative (D1) method achieved the highest accuracy of 67.15% during testing. The k-nearest neighbor (KNN) algorithm delivered superior performance with a test accuracy of 86.35%. Competitive adaptive reweighted sampling (CARS) proved to be the most effective dimensionality reduction tool, maintaining a test accuracy of 96.77%. Among the optimization techniques, the Goose Optimization (GO) algorithm yielded the most robust results, achieving 100% accuracy across trainingand test metrics. The integration of portable NIRS with the LOF-D1-CARS-GO-KNN strategy presents a powerful, scientifically reliable approach for tracing the geographical origin of turmeric, thereby ensuring product authenticity and quality. • Proposed LOF-D1-CARS-GO-KNN algorithm for rapid turmeric origin dentification using portable NIR + chemometrics. • GO algorithm outperforms GWO/PSO in model optimization for turmeric identification. • The method achieves high accuracy and efficiency with great practical application potential.