Fig. 2(b) exhibits the relationship between the loadings of PC1
and PC2. These loadings express how important each bin is when
determining the value of the component in the light of a given
sample. One can see that on the one hand most of the loadings are
around zero and, on the other hand, that the bins d 1.27 and
1.62 ppm are the most important for variables PC1 and PC2,
respectively. Besides, Fig. 2(b) also indicates that bins d 1.62 and
1.57 ppm are the best discriminatory ones since they are the most
distant from zero for both axes. In fact, this reasoning is also
Fig. 2. PCA scores (a) and loadings (b) scatter plot of PC1 vs PC2 for discriminating analysis between non-irradiated (blue) and irradiated doses of 1 kGy (red) and 5 kGy (green)
soybeans.
Fig. 3. PCA scores scatter plot of PC1 vs PC2 for discriminating among cultivars.
Samples under (above) the dashed line are from the irradiated (non-irradiated)
group.
Fig. 4. PLS scores scatter plot (component 1 vs component 2) discriminating between
irradiated (red) and non-irradiated (green) soybeans. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this
article.)
A.S. Ribeiro et al. / Food Control 36 (2014) 266e272 269
confi rmed by hypothesis tests such as that proposed by Brunner
and Munzel (2000), where the p-value for differences between
the means of the intensity in the bins d 1.62 and 1.57 ppm for
irradiated and non-irradiated groups is less than 1$10 10. This
conclusion is also obtained by VIP score study as can be seen later in
this section.