[zo7]

One way to do PCA is using SVD to find a transformation matrix of eigenvectors to project your data with, so they're similar.