Empirical approach to improve the prediction of soluble solids content in mango using near-infrared spectroscopy
2020
Phuangsombut, K. | Phuangsombut, A. | Terdwongworakul, A.
The present work studied the effect of peel on the performance of near-infrared (NIR) spectroscopy for estimating the soluble solids content (SSC) of mangoes and on SSC estimation improvement using the empirical approach developed from our previous work. The results showed that a partial least squares (PLS) regression model based on intact fruit or unpeeled NIR spectra had lower accuracy (correlation coefficient, r = 0.84; root mean squared error of prediction, RMSEP = 1.50 °Brix) than using flesh or peeled fruit NIR spectra (r = 0.88; RMSEP = 1.27 °Brix). Improved prediction was obtained (r = 0.87 and RMSEP = 1.36 °Brix) by applying the empirical approach (which partially removed the difference between the flesh and the peel, and the flesh spectra) to the original peel and flesh spectrum prior to developing the PLS model. The empirical approach proved beneficial in improving the prediction of the SSC in mango fruit which was affected by the peel.
显示更多 [+] 显示较少 [-]