Electronic Nose for Tea Identification Detection based on Machine Learning K-Nearest Neighbors Method and Raspberry Pi 4

This study aims to identify the types of tea to determine their authenticity and quality using the Machine Learning K-Nearest Neighbors (KNN) method and Raspberry Pi 4. The developed system uses tea aroma data collected through an Electronic Nose (E-Nose) connected to Raspberry Pi 4, which utilizes eight gas sensors (MQ 2, MQ 3, MQ 4, MQ 6, MQ 7, MQ 8, MQ 9, MQ 135) to measure the gas levels formed by various volatile compounds that evaporate in different types of tea with varying compositions, namely 100% green tea, 100% jasmine tea, 100% black tea, 80% green tea and 20% jasmine tea, and 75% green tea and 25% jasmine tea. The data is processed using the Machine Learning KNN method to classify tea types based on the detected aroma patterns. The evaluation of the analysis takes into account metrics such as accuracy, precision, recall (sensitivity), true negative rate (specificity), F-1 score, confusion matrix, and Principal Component Analysis (PCA). The results of this study indicate that the Machine Learning KNN method using Raspberry Pi 4 can accurately identify the types of tea. The evaluation of the analysis shows satisfactory model performance, with high levels of accuracy, precision, recall, specificity, and F1 scores. The confusion matrix provides a clear picture of the model's ability to classify tea types, while the PCA plot provides an intuitive understanding of the data structure, making it easier for researchers and stakeholders to interpret and analyze the complexity of the data with ease. The results of this study show that the E-Nose system with the KNN method is capable of differentiating between green tea, black tea, and a combination of green tea and black tea, with an accuracy value of up to 93%.