LAUSR

Contemporary consumers drink significant amounts of tea because of its health–benefits mainly associated to the presence of polyphenols with high antioxidant activity. Therefore, the information how to obtain tea infusion of the highest quality, i.e. with high antioxidant capacity is needed. In this study, the various models for the prediction of total polyphenols and antioxidant activity of green and black tea infusions were developed and compared. Three mathematical equations: Spiro’s, Peleg’s and logarithmic, and two data mining techniques: multivariate adaptive regression splines (MARS) and artificial neural networks (ANNs) were used to build the predictive models. The results obtained show that Spiro’s model could be used for the prediction of green tea quality expressed as total phenolic content or the antioxidant activity (determination coefficients above 0.99), whereas Peleg’s model is more suitable for black tea quality prediction (determination coefficients above 0.99). Data mining techniques (MARS and ANNs) enable to create models fast and of simple application with very good acceptability (determination coefficients above 0.99).