-----------------------------------------------------------
Statistics Seminar Series
Thursday, October 4, 2007 @ 4:00PM
Cullimore Hall, Room 611
New Jersey Institute of Technology
Randomly Trimmed Mean
Zhiqiang Peter Chen, PhD.
Department of Mathematics
William Paterson University
Abstract
Let μ, and S be robust location and scale estimator respectively and let c be any tuning constant, we trim off the observations that are outside the interval of [μ-cS, μ+cS]. Randomly trimmed mean is defined by taking average of the remaining data. We will present theoretical results of randomly trimmed means in term of robustness and asymptotic theory.