Principal Component Analysis

A linear dimensionality-reduction technique that identifies orthogonal directions (principal components) in data which successively capture maximal variance; typically computed via diagonalization of the empirical covariance matrix or via singular value decomposition of the data matrix.