University of Waikato 2006-05-23 I1.05 (I Block, 1st floor) University of Waikato Abstract: In the classical model, we have standard method such as MLE to estimate the parameters, but in the robust literature, this concept is lacking. We are trying to establish unified approach to estimate the parameters for the contaminated data. Here we employed maximum likelihood estimation procedure, which is very sensitive to the outliers. In order to avoid this situation, we add another parameter free known function to the original distribution and employ EM algorithm, which gives less weight for the far away observations from mean, to estimate the parameters. We refer this procedure as MixM-estimators. These estimates compare with standard robust estimators such as Huber estimate and Tukey estimate. Influence functions for these robust estimates are computed. From the simulation study, we can say it is bounded. In this talk, I will give general approach, which can explain with particular problem of robust estimation of location parameter. |
