Abstract:
In this paper we prove the consistency in probability of a class of generalized BIC criteria for model selection in nonlinear regression, by using asymptotic results of Gallant. This extends a result obtained by Nishii for model selection in linear regression.