> X<-yarn$NIR
> X_nipals<-nipals(X,a=10,it=100)
Two matrices are generated (P and T)
As in other posts, we are going to look to the loadings & scores, for firsts three principal components:
> wavelengths<-seq(1,268,by=1)
> matplot(wavelengths,X_nipals$P[,1:3],lty=1,
+ pch=21,xlab="data_points",ylab="log(1/R)")
> T3cp<-X_nipals$T[,1:3]
> pairs(T3cp)
In the following plot, I compare the loadings plots for the first 3 PCs calculated with SVD (up) and with NIPALS (down):We can see how the first PC (PC1) has the same shape for both. The other two (PC2 & PC3) has also the same shape, but inverted).
Let´s compare the scores plots:
Red dots for NIPALS, black for SVD:
No hay comentarios:
Publicar un comentario