The purpose of this study is to investigate the feasibility of discriminating the different varieties, production areas and seasons
of Taiwan partially fermented tea by using Near Infrared Spectroscopy (NIRS). A total of 308 partially fermented tea samples with 6 different tea varieties, 6 production areas and 2 different production seasons were collected and analyzed. The principal component analysis (PCA) result of NIRS spectra data showed that the first three principal components could explain the sample variation up to 95.0%. The ability of classifying different production areas of tea samples by PCA was the best followed by tea varieties. The discriminant model further established by NIRS data with partial least square (PLS) could recognize and identify the varieties, production areas and seasons of tea samples up to 98.4% (299 of 305), 97.4% (296 of 304), and 100%, respectively. Using the
established discriminant model, the tea samples with different varieties, production areas and seasons could be correctly predicted
and identified at the levels of 96.3%, 94.1% and 99.2%, respectively.