SILPAKORN UNIVERSITY SCIENCE AND TECHNOLOGY JOURNAL

Volume 14, No. 03, Month SEPTEMBER, Year 2020, Pages 203 - 214

Liu-type logistic regression coefficient estimation with multicollinearity using the bootstrapping method

Narumol Sudjai, Monthira Duangsaphon

Abstract Download PDF

This study proposed new estimators for shrinkage and ridge parameters to overcome the multicollinearity problem in Liu-type logistic regression using the bootstrapping method. Moreover, we compared the performance of four methods for logistic regression coefficient estimation with multicollinearity present: the maximum likelihood estimator, ridge logistic regression, Liu logistic regression, and Liu-type logistic regression, all performed with the bootstrapping method. A simulation study was conducted to compare the performance of the four different estimation methods using the estimated mean square error. The results from both the simulation study and a real data application showed that the Liu-type logistic regression with the bootstrapping method performed best, among the four methods, with a high correlation coefficient. Moreover, the proposed estimators for the shrinkage parameter and ridge parameter showed good performance. In addition, the use of Liu-type logistic regression together using the bootstrapping method was the most robust for correcting the multicollinearity problem.

Keywords

bootstrapping method, ridge estimator, Liu estimator, Liu-type estimator, multicollinearity

---