The authentication of the local cured Nicotiana tabacum L. based on varieties by using machine learning algorithms

• The sensorial quality of each tobacco variety is different according to its grade • Machine Learning was used for non-destructive to discriminate tobacco varieties • Logistic Regression and Neural Network showed the highest accuracy • Classification was achieved with an overall AUC for the cross-validation equal to 1 Tobacco ( Nicotiana tabacum L.) is a valuable crop with great economic and scientific importance and some important varieties are cultivated in different regions of Indonesia. With the rise in tobacco smuggling and counterfeiting, there is a need for robust methods to authenticate the origin of tobacco leaves. Existing analytical approaches provide high accuracy but require expensive instruments, complex sample preparation, and laboratory expertise, limiting their practicality for routine industrial use. This study evaluated the potential of machine learning (ML) techniques coupled with simple color image analysis as a rapid, cost-effective alternative to accurately classify processed tobacco's origins. Our research demonstrated that the grade A of varieties Garut, Rengel, Tambeng and Gayo possess a higher overall acceptance compared to the other varieties based on sensory analysis. Moreover, the ML algorithms develop a reliable classification system that can identify unique characteristics within color images corresponding to 6 different tobacco varieties (6 classes) from different origins in Indonesia as well as the classification of their 2 grades on those 4 varieties (8 classes). Several algorithms such as logistic regression, random forest, and support vector machine, have been tested, and all of them have shown excellent results approaching 100% accuracy in calibration and prediction. These results show that ML models built on simple digital images can address the limitations of traditional analytical methods. This approach offers a practical way to verify tobacco variety and origin, helping improve quality control and reduce fraud in the tobacco industry.