QR decomposition¶

\[X = QR\]

  • Q is orthonormal

  • R is upper triangular

QR decomposition is numerically very stable at it takes \(O(ND^2)\)

SVM for regression. Kernels for document comparison