Geoscience Reference
In-Depth Information
Fig. 9.3 Principal component loads suggesting that the PCs are inl uenced by dif erent
minerals. See text for detailed interpretation of the PCs.
percent_explained =
72.4362
22.7174
4.8463
We see that more than 72% of the total variance is contained in PC 1 , and
about 22% is contained in PC 2 , while PC 3 contributes very little to the total
variance of the data set (~5%). h is means that most of the variability in the
data set can be described by just two new variables. As would be expected,
the two new variables do not correlate with each other, as illustrated by a
correlation coei cient between newx(:,1) and newx(:,2) that is close to zero.
corrcoef(newx(:,1),newx(:,2))
ans =
1.0000 0.0000
0.0000 1.0000
We can therefore plot the time series of the thirty samples as two independent
variables, PC 1 and PC 2 , in a single plot.
plot(1:30,newx(:,1),1:30,newx(:,2))
grid, legend('PC1','PC2')
xlabel('Sample ID'), ylabel('Value')
h is plot displays approximately 94% (72%+22%) of the variance contained
in the multivariate data set. According to our interpretation of PC 1 and
Search WWH ::




Custom Search