In relation to objective colour measurements, although the variance
explained by the model was higher than 80% for L* value, the root mean
square error of cross-validation was still high (RMSECV= 1.27) as compared
to SD, resulting in a ratio performance deviation (RPD = 1.69)
not high enough to meet the requirements for analytical purposes
(Williams, 2001, 2008). NIRS predictability was lower for a* and b*