로지스틱회귀모형의 로버스트 추정을 위한 알고리즘

Abstract

The maximum likelihood estimation is not robust against outliers in the logisticregression. Thus we propose an algorithm for the robust estimation, which identies thebad leverage points and vertical outliers by the V-mask type criterion, and then strivesto dampen the eect of outliers. Our main nding is that, by an appropriate selectionof weights and factors, we could obtain the logistic estimates with high breakdownpoint. The proposed algorithm is evaluated by means of the correct classicationrate on the basis of real-life and articial data sets. The results indicate that theproposed algorithm is superior to the maximum likelihood estimation in terms of theclassication.