Median-of-means Estimation Of Parameters In Logistic Regression Model

In this paper,the Median-of-Means method is introduced into the Logistic regression model to estimate unknown parameters,and a new estimator is proposed.This estimator is a robust estimation.We further prove the consistency and asymptoticity of the estimator normality.This article considers multiple Logistic regression models,focusing on the situation when outliers appear in the model.Numerical simulation results show that our proposed The value of the estimator is closer to the true value than the traditional estimator.We will also this method is applied to breast cancer diagnosis data,credit score data and medical staff depression questionnaire survey data.In actual data analysis,we use KS value and ROC curve to evaluate the stability of the model,and further more performance of Median-of-Means method by the accuracy and sensitivity.