Abstract:
In this paper, an improved ridge type estimator is introduced to overcome the effect of multicollinearity
in logistic regression. The proposed estimator is called a modified almost unbiased
ridge logistic estimator. It is obtained by combining the ridge estimator and the almost
unbiased ridge estimator. In order to asses the superiority of the proposed estimator over the
existing estimators, theoretical comparisons based on the mean square error and the scalar
mean square error criterion are presented. A Monte Carlo simulation study is carried out to
compare the performance of the proposed estimator with the existing ones. Finally, a real
data example is provided to support the findings.
Key words: Logistic Regression, Multicollinearity, ridge estimator, Modified almost unbiased
ridge logistic estimator, Mean square error.
