The discriminant analysis demonstrated that mango fruit can be
classified into three DAFS-assigned classes of maturity based on
non-destructive predictor variables. SG was shown to be the best
contributing variable for the classification. Cluster analysis was
executed to allocate each individual mango into four maturity classes with the employment of TSS, TA and DAFS as categorization
parameters. With variable selection by means of a stepwise method, simplification of the model with a reduced number of the
non-destructive variables could be achieved. The performance of
the simplified model for the classification of four maturity levels
excelled with an accuracy of 89.0% which was deemed appropriate
for practical application. However, the validation of the model with
a separate set of mango samples in later season needs to be further
investigated. Finally, for the maturity classification of mangoes cv.
Nam Dokmai, the best applicable variables of physical, mechanical,
and optical properties were SG, stiffness coefficient, and diffuse
reflectance at 670 nm, respectively