RAPID DETERMINATION OF MATURITY IN APPLE USING OUTLIER DETECTION AND CALIBRATION MODEL OPTIMIZATION

A technique to predict the maturity quality of intact apple fruit measured non-destructively by Fourier transform near-infrared (FT-NIR) spectroscopy in the wavelength range of 814-1100 nm was investigated. Mathematical models for calibration and prediction of sugar content (SC) and titratable acidity (TA) indices of maturity were developed by partial least squares (PLS) regression. The modeling procedures were systematically studied with the focus on outlier detection and calibration model optimi‘zation. By using two outlier detection techniques, 321 optimal sample sets were successfully chosen from the original 333 sample sets. The optimization regression models for maturity were obtained in the wavelength range of 814-1100 nm with correlation coefficients of 0.95 and 0.74 and standard errors of prediction of 0.54 and 0.04 for SC and TA, respectively. The results of this method for the determination of maturity were compared with those of reference methods, with no significant difference at the 0.05 level. It was demonstrated that outlier detection methods were very helpful for optimizing FT-NIR calibration models and would improve the accuracy of the prediction models.