This is a preview. Log in through your library . Abstract We introduce a randomized algorithm for overdetermined linear least-squares regression. Given an arbitrary full-rank m × n matrix A with m ≥ n ...