Convenient and accurate method for the identification of Chinese teas by an electronic nose

A convenient, accurate, and effective approach for the identification of Chinese teas and their production area has been developed. For this, Chinese tea samples from different regions were collected and their odours were analysed by an electronic nose (E-nose). An unambiguous identification of the Chinese teas could not be achieved by means of traditional principal component analysis or linear discriminant analysis methods. Thus, multiple logistic regression (MLR), support vector machines (SVM), and random forests (RF) were employed as alternative to build identification models. The experimental results show that the method aiming within scope based on the RF performs very well, with prediction accuracies and computation times being superior to the two others (MLR and SVM). The results were demonstrated that E-nose could be used in the classification of Chinese teas, when an optimal pattern recognition algorithm is selected. The present study provides a critical outlook on the developments of Chinese teas identification, authenticity control and against adulteration in the Chinese circulation market.