Statistics Colloquium
University of Connecticut
Storrs, Connecticut
The Department of Statistics Cordially invites you to a Colloquium
Lixing Zhu
Chair Professor and Department Head
Department of Mathematics
Hong Kong Baptist University
Asymptotic composite regression
ABSTRACT
Composition methodologies in the literature have been applied to variance reduction via the direct linear combination of either initial estimators or objective functions. Unlike these methodologies, the asymptotic presentation of initial estimators and its relationship to model-independent parameter values are used to propose a novel approach. A least squares fitting can then be applied to optimize the weights in the composition such that both variance and bias reduction can be achieved. The examples are quantile regression and blockwise empirical likelihood, which have a smaller limiting variance; the Stein estimator, which achieves both smaller bias and smaller variance; and nonparametric kernel estimation which has a faster convergence rate than the classical optimal one. Simulations are conducted to examine its performance in finite sample situations and a real dataset is analysed for illustration.
DATE: Wednesday, September 17, 2014
TIME: 4:00 p.m.
PLACE: Philip E. Austin Building – Room 105
Coffee will be served at 3:30 in room 326
For more information, contact: Tracy Burke at tracy.burke@uconn.edu