Electronic Nose for Organoleptic Score Prediction of Gambung Green Tea Using Optimized Regression Algorithms

The application of Electronic Nose (E-nose) technology using a system to evaluate the aroma-based green tea quality has gained significant attention in the tea industry. This study aims to predict the organoleptic score of Gambung green tea using gas sensor data MQ3, MQ5, MQ138, TGS822, TGS2602, TGS2620 analyzed with regression algorithms. Five regression models were evaluated Decision Tree, K-Nearest Neighbors, Random Forest, XGBoost, and AdaBoost. Experimental results demonstrate that Random Forest attained optimal performance, as shown by an R-squared value of 0.9857 and a Mean Squared Error of 0.0854, followed by KNN R-squared 0.9849 and MSE 0.0903, XGBoost with R-squared 0.9832 and MSE 0.1005. AdaBoost exhibited the least performance with an R-squared score of 0.8706 and MSE of 0.7754. This study supports the development of tea quality assessment techniques by integrating E-nose sensors with machine learning approaches, offering practical relevance for food technology and agribusiness industries.