Verification by Monte Carlo simulation of the efficiency of application of the likelyhood ratio test for identity of non-linear models / VerificaÃÃo por simulaÃÃo monte carlo da eficiÃncia da aplicaÃÃo do teste de razÃo de verossimilhanÃa para identidade de modelos nÃo lineares

AUTOR(ES)
DATA DE PUBLICAÇÃO

2006

RESUMO

An alternative to compare non-linear models is given by means of the test of identity of models. The statistics utilized normally is given by means of the likelihood ratio. Aiming to evaluate by means of the Monte Carlo simulation the performance of this test, by the aid of the different sample models, number of equations, robustness and different models, the control of the type I error and the power was considered as a criterion . it follows that the increase in the number of equations produced a conservative effect in the control of the type I error for the Gompertz and Langmuir models for all the sample sizes and that the Gompertz model proved more powerful and robust for non-normal residuo

ASSUNTO(S)

regressÃo nÃo-linear comparaÃÃo de modelos simulaÃÃo identity of non-linear models monte carlo simulation non-linear regression estatistica

Documentos Relacionados