Robust Estimators of Logistic Regression with Problems Multicollinearity or Outliers Values.

Abstract

Whenever there is a relationship between the explanatory variables (X_S). This relationship causes multicollinearity which in turn leads to inaccurate and bias estimations of the model parameters. Therefore, this results in high discrepancy that influences the next phase of the statistical inference where (OLS), method loses its features having the lowest variance. Consequently, this paper concerns itself with figuring out methods that can be applied by researchers and those who are interested in this field to overcome this problem using (Ridge) method. Moreover, the paper seeks to solve other problems such as the loss of normal distribution property or abnormalility by means of methodical means including (Ridge) and (Robust Ridge). However this study is applied through simulation experiments aim at producing the data of the model. Based on these experiments and tests, the research has come up with the result that (Robust Ridge) is the best method that might be employed to solve the problem of has both normal and abnormal data for the estimation of the parameters of the Logistic Regression Model.