Abstract

Translucent flesh in mangosteens has been studied extensively in order to predict its presence in the fruit non-destructively. In this work, we propose the use of electrical parameters, capacitance and resistance of mangosteens, to determine the defect. The data are then analyzed employing logistic regression analysis to predict the probability for classification groups; in this case, those with and without translucent defect. The result from the logistic regression analysis reveals that the overall correctness of the prediction at 0. 5 cut point is 87. 3% which is competitive to those obtained by other means. © 2013 The Korean Society of Food Science and Technology and Springer Science+Business Media Dordrecht.