Evaluation of the composition and sensory properties of tea using near infrared spectroscopy and principal component analysis

This paper presents a thorough study of the methods available for evaluating tea quality by using near infrared (NIR) spectroscopy technology as an objective universal sensory assessment procedure. The first section of this systematic study reports the assessment of tea quality by correlating chemical components with sensory evaluation using principal component analysis scores and loadings. The first three PCs account for more than 94% of the variance in input data for four original chemicals (lignin, polyphenols, caffeine and amino acids). The correlation between PC1 scores alone and tea quality classes, given by sensory tastes, is higher with R-2 > 0.98 in green and black teas. The second section focuses on the preparation of multi-equations in order to calibrate NIR determination of chemical components according to sensory assessment. The outcome of the established multiple linear regression equations was an ability to predict tea quality directly by NIR spectroscopy with a correlation r(2)=0.86 and with less than 9.7% misclassified samples in green and black tea. In the final section, some poor correlations between tea quality and market prices in the world are reported and we suggest that NIR technology can provide an internationally accepted, objective and rapid method to assess tea quality.