In this paper, in order to test whether changes have occurred in a nonlinearparametric regression, we propose a nonparametric method based on the empiricallikelihood. Firstly, we test the null hypothesis of no-change against thealternative of one change in the regression parameters. Under null hypothesis,the consistency and the convergence rate of the regression parameter estimatorsare proved. The asymptotic distribution of the test statistic under the nullhypothesis is obtained, which allows to find the asymptotic critical value. Onthe other hand, we prove that the proposed test statistic has the asymptoticpower equal to 1. These theoretical results allows find a simple teststatistic, very useful for applications. The epidemic model, a particular modelwith two change-points under the alternative hypothesis, is also studied.Numerical studies by Monte-Carlo simulations show the performance of theproposed test statistic, compared to an existing method in literature.
展开▼