14th IFAC Symposium on System Identification, SYSID 2006

SYSID-2006 Paper Abstract


Paper ThB1.5

De brabanter, Jos (Katholieke Univ. Leuven), Pelckmans, Kristiaan (Katholieke Univ. Leuven), Suykens, Johan (K.U. Leuven), De Moor, Bart (Katholieke Univ. Leuven)

Generalized Likelihood Ratio Statistics Based on Bootstrap Techniques for Autoregressive Models

Scheduled for presentation during the Invited Session "Nonlinear System Identification II" (ThB1), Thursday, March 30, 2006, 16:50−17:10, Concert Hall

14th IFAC Symposium on System Identification, March 29 - 31, 2006, Newcastle, Australia

This information is tentative and subject to change. Compiled on July 22, 2018

Keywords Nonparametric Methods, Maximum Likelihood Methods, Model Validation


Parametric models provide explanatory power and can give a parsimonious description for dynamical systems modelling. But, there is a risk thatmisspecication of an underlying stochastic model can lead to misunderstanding of the system, wrong conclusions, and erroneous predicting. It is common practice to check whether a parametric model fits a given data set reasonably well. To achieve this, in the Neyman-Pearson framework, one needs to specify a class of alternative(nonparametric models). In this paper, we use the idea of the generalized likelihood ratio test based on bootstrap techniques. This technique can be applied for each parametric model structure and its nonparametric counterpart. We demonstrate the use of these methods for testing the significance of NARX versus ARX models.