论文信息 - A simulation study of structure characterization methods

A simulation study of structure characterization methods

Linear black box modelling often requires an inductive model structure characterization method or expressed in other terms a technique that permits for the determination of the order of the model on the basis of experimental data. Recently a number of different methods like FPE, AIC, LILC and BIC have been proposed. The statistical properties of those criteria are badly known. In the paper, the criteria are shown to be similar, differing only in a threshold level. A systematic Monte-Carlo simulation study with second order autoregressive models has been carried out. It is found that the properties of the structure discriminating statistics are remarkably independent of the parameter values or the location of the poles. For all criteria, the error rate increases as the coefficient of the highest order time lag decreases to zero. The simulation indicates the superiority of BIC over LILC and of LILC over AIC. The study suggests that a new criterium that would be dependent on the highest order time lag parameter could be superior.

J. A. Spriet | J. Spriet | P. Herman | P. Herman

[1] Robert B. Ash,et al. Information Theory , 2020, The SAGE International Encyclopedia of Mass Media and Society.

[2] R. Kashyap,et al. Dynamic Stochastic Models from Empirical Data. , 1977 .

[3] H. Akaike. Statistical predictor identification , 1970 .

[4] B. G. Quinn,et al. The determination of the order of an autoregression , 1979 .

[5] R. Kashyap. Inconsistency of the AIC rule for estimating the order of autoregressive models , 1980 .

[6] R. Bhansali,et al. Some properties of the order of an autoregressive model selected by a generalization of Akaike∘s EPF criterion , 1977 .

[7] R. Kashyap. A Bayesian comparison of different classes of dynamic models using empirical data , 1977 .

[8] H. Akaike. A Bayesian extension of the minimum AIC procedure of autoregressive model fitting , 1979 .

[9] H. Akaike,et al. Information Theory and an Extension of the Maximum Likelihood Principle , 1973 .

[10] Hirotugu Akaike,et al. On entropy maximization principle , 1977 .

[11] R. Shibata. Selection of the order of an autoregressive model by Akaike's information criterion , 1976 .

[12] J. Makhoul,et al. Linear prediction: A tutorial review , 1975, Proceedings of the IEEE.

[13] H. Akaike. Canonical Correlation Analysis of Time Series and the Use of an Information Criterion , 1976 .