Parameter Estimation Of Poisson Regression Model Based On Median-of-Means

In this paper,we introduce the Median-of-Means estimator of Poisson regression model.The consistency and asymptotic normality of the estimator are proved.We set up several Poisson regression models,using numerical simulation under the condition of no outliers and outliers to compare the performance of the maximum likelihood estimator and the Median-of-Means estimator.The simulation results show that the Median-ofMeans method is more robust.We also apply the maximum likelihood method and the Median-of-Means method to two actual data sets: the number of coal seam fractures and the number of damaged positions on the fuselage,then the k-fold cross validation method is used to demonstrate the advantages of our method.