FINITE SAMPLE AIC FOR AUTOREGRESSIVE MODEL ORDER SELECTION

An estimate for the prediction error of the least-squares-forward (LSF) autoregressive (AR) parameter estimation method has been recently proposed. In this paper, this estimate is used for deriving a new AR model order selection criterion. This new criterion is an estimate of the Kullback-Leibler index and can replace the Akaike information criterion (AIC) and its corrected version AICC. In a simulation study, the performance of this new criterion and other existing order selection criteria is examined in the finite sample case. Simulation results show that the performance of the proposed criterion is much better than the other theoretically derived criteria.