[Statlist] CANCELLED: Research Seminar in Statistics *FRIDAY, 13 MARCH 2020* GSEM, University of Geneva

[gsem-support-instituts]

Dear All,

Due to the evolving situation with COVID-19, the RCS has decided to cancel the seminar from tomorrow. 

Thank you very much for your understanding.



Sandra Vuadens 
Assistant to research and institutes
gsem-support-instituts using unige.ch





-----Message d'origine-----
De : gsem-support-instituts 
Envoyé : lundi, 9 mars 2020 08:59
Cc : Sebastian Engelke <Sebastian.Engelke using unige.ch>
Objet : Research Seminar in Statistics *FRIDAY, 13 MARCH 2020* GSEM, University of Geneva
Importance : Haute

Dear All,

We are pleased to invite you to our next Research Seminar.

Looking forward to seeing you


Organizers : E. Cantoni - S. Engelke - D. La Vecchia - E. Ronchetti S. Sperlich - F. Trojani - M.-P. Victoria-Feser


FRIDAY, 13 MARCH 2020 at 11:15am, Uni-Mail M 5220

Asymptotically Optimal Bias Reduction for Parametric Models Mucyo KAREMERA - University of Geneva, Geneva School of Economics and Management

ABSTRACT:
An important challenge in statistical analysis concerns the control of the finite sample bias of estimators. This problem is magnified in high dimensional settings where the number of variables p diverges with the sample size n, as well as for nonlinear models and/or models with discrete data. For these complex settings, we propose to use a general simulation-based approach and show that the resulting estimator has a bias of order O(0), hence providing an asymptotically optimal bias reduction. It is based on an initial estimator that can be slightly (asymptotically) biased, making the approach very generally applicable, especially in complex settings where classical estimators, such as the Maximum Likelihood Estimator (MLE), can only be (numerically) approximated. We also derive its asymptotic properties which show that a trade-off can be sought between efficiency and computational cost. Moreover, this optimal bias reduced estimator does not rely on model-based analytical transformations and is readily applicable. We show that the Iterative Bootstrap of A.Y.C. Kuk (1995) provides a computationally efficient approach to compute it. Using this framework, we develop new optimal bias reduced estimators for logistic regression models, with and without random effects, that enjoy additional properties such as robustness to data contamination and to the problem of separability.


Visit the website: https://www.unige.ch/gsem/en/research/seminars/rcs/