Performing principal components analysis on the N1-by-N2 real-valued data matrix X, where N1 and N2 are the number of features (N1 variables) and observations (N2 samples), respectively.