Prediction of overall sweet, overall sour and firmness of mango by near infrared spectroscopy.
Klamcharoen O., Suwonsichon T., Suwonsichon S., Ritthiruangdej P., Kasemsumran S.
Author Affiliation: Department of Product Development, Faculty of Agro-Industry, Kasetsart University, Bangkok 10900, Thailand.
: P96
Abstract : The objective of this study was to apply near infrared spectroscopy (NIRS) for predicting sensory attributes of 2 Thai mango varieties (Okrong and Numdokmai). Two hundred samples of each variety were divided into 2 sets as calibration set (n=133) and validation set (n=67). All samples were determined by NIRS in the wavelength region of 1100-2500 nm. All spectra were pretreated by the 2nd derivative before developing model. After NIRS measurements, all samples were evaluated sensory descriptive analysis. Eight trained panelists evaluated samples and rated their intensities of overall sweet, overall sour and firmness. To develop the predicting models, data analysis of calibration set were done by the partial least square regression (PLSR). Result showed that the PLSR models provide Rcal in the range of 0.800 to 0.893. PLSR models were also applied to find out their performances in validation data set. Results showed that all PLSR models were applicable. They provided the standard error of prediction (SEP) lower than 1.161 and the ratio of standard error of prediction to standard deviation of prediction (RPD) greater than 2.06.