FAST SUCCESSIVE SELECTION OF VARIABLES IN LINEAR-MODELS USING MODIFIED QR FACTORIZATION

A new approach for succesively selecting regressor variables in the li near-in-the-parameter modelling problem is presented. It is based on Q R with column pivoting factorisation, where the columns are pivoted ba sed on the correlation with the corresponding rotated output vector. I n contrast to other competing methods, no parameter estimation is nece ssary. The method is computationally fast and numerically robust.