Abstract
To apply the partial least square regression (PLS) to the growth diagnoses on the fruit of Japanese pear by Near infrared spectroscopy (NIRS), measurements for constituent sugar concentrations (%w/w) in the juice of the developing Japanese pear fruits by NIRS were conducted by PLS, and the assignment of PLS latent variables as independent variables were examined by using PLS loadings.
Results of the PLS regression indicated that the optimum number of scores for sucrose, glucose, fructose, and sorbitol were 10, 5, 11, and 10. The correlation coefficients (R) were 0.96, 0.90, 0.99 and 0.99, with the standard error of prediction of 0.21, 0.17, 0.19 and 0.31 respectively.
The first calculated latent variable of each constituent sugar appeared to integrate water information. In the second and subsequent loadings, many absorption bands assigned to sugar were loaded with sugar information. It indicated that PLS regression was effective analytical method for the growth diagnoses on Japanese pear fruits by NIRS.
