[vecter]

How is PCA the same as least squares? I've always understood it as the eigendecomposition of the covariance matrix.