School Seminars and Colloquia

M-Estimation and Its Recursive Analog

Applied Statistics Seminar

by Amy Wu

Institution: York University, Canada
Date: Thu 8th March 2007
Time: 12:00 PM
Location: Room 213, Richard Berry Building, The University of Melbourne

Abstract: In this talk, some M-estimation based procedures are reviewed. It is noted that many M-estimates have no explicit expressions. Often the Newton approach could not be applied to the computation of M-estimates. Even when the Newton approach is applicable, it is usually selective for the initial values. Thus, when a new observation is obtained, it is onerous and gaucherie to recalculate the estimates from enlarged data set. Hence, there is a need to construct a convenient algorithm up-dating the M-estimates when new
observations are consecutively obtained and the estimates need to update from time to time. Some up-dating recursion formulas of M-estimation will be presented in this talk.

For More Information: Dr Owen Jones: