Order selection for AR models by predictive least-squares
暂无分享,去创建一个
We present a new criterion for selecting the order of AR models which, unlike the existing criteria, is amenable to on-line or adaptive operation. It is based on the Predictive Least-Squares principle and is implemented in a computationally efficient way by predictive lattice filters. We prove the consistency of the criterion and demonstrate its performance by computer simulations.