Classification of Lu'an Gua Pian tea before and after Qingming Festival using HPLC-DAD analysis: a comparison of different data analysis strategies

Abstract

Lu'an Gua Pian tea (LAGP) is a traditional Chinese historical tea and one of the top ten famous teas in China. The price of LAGP from the same place of origin varies greatly in the market depending on the harvest time, with the LAGP harvested before the Qingming Festival (between April 4th and 6th every year) being regarded as a precious product. To accurately determine the harvest time of LAGP, especially around Qingming, HPLC-DAD combined with three different data analysis strategies (including targeted component analysis, non-targeted component analysis, and non-targeted fingerprint analysis) was evaluated and compared. Four machine learning algorithms were used to build the corresponding classification models, among which principal component analysis-linear discriminant analysis (PCA-LDA), partial least squares discriminant analysis (PLS-DA) and K-nearest neighbors (K-NN) achieved more than 95% classification accuracy. In addition, the non-targeted component analysis can classify LAGP picking time to a much smaller extent with an accuracy of 100% for PLS-DA. The advantages and disadvantages of the three data analysis strategies were compared. Regardless of the data analysis strategy used, the final classification accuracy was satisfactory. The appropriate data analysis strategy can be selected according to the specific experimental purpose. This work provides a variety of alternative solutions based on HPLC-DAD for identifying the picking time of Lu'an Gua Pian.