Tidyverse and Chemometrics (part 11): Proyecting Fish2 on Fish1 PCs

Once I have the PCs for the 40 fish meal samples calculated on the previous post is time to project the new samples on the fish1 PCs in order to see which samples will fill and expand the space.

 

First we have to calculate the fish2 scores on the fish1 principal components space multiplying the X matrix of fish1 by the loadings matrix "P" of fish1 PCs, and the results are the scores of fish2 on fish1 PCs.

 

Now if we use the drawMahal from the Chemometrics Package we get these plots which show that we need to send some of the fish2 samples to the lab to expand the calibration.

As we can see samples goes apart specially on the third PC. What you think can be the reason?.

PC1 is quite related to moisture and it does seem that this constituent is not two much expanded with fish2 spectra, but which constituent is specially represented by the third loading of the fish1 PCs loadings?.