Estimations and Optimal Tests in Some Parametric Models
Résumé
In the present paper, we introduce an efficient method for the estimation in the multidimensional case. The
key idea is based on a good assessment of the error without using confidence intervals. The consistency of the proposed
estimate is established. Consequently, we discuss the estimation in statistical tests corresponding to parametric context,
and prove that this kind of estimators ensures the optimality of statistical tests. We partially extend the scope of our study
to some processes. In order to examine the performance of our methodology, finite sample results are performed. This
work completes and extends in nontrivial way the results obtained by Lounis (2017).