期刊名称：International Journal of Statistics and Applications
印刷版ISSN：2168-5193
电子版ISSN：2168-5215
出版年度：2020
卷号：10
期号：3
页码：55-59
DOI：10.5923/j.statistics.20201003.01
语种：English
出版社：Scientific & Academic Publishing Co.
摘要：The aim of our study is to select the best method for overcoming partial and full multicollinearity in binary logistic model for different sample sizes. Logistic ridge regression (LRR), least absolute shrinkage and selection operator (LASSO) and principal component logistic regression (PCLR) compared to maximum likelihood estimator (MLE) using simulation data with different level of multicollinearity and different sample sizes (n=20, 50, 100, 200). The best method is chosen based on mean square error (MSE) values and the best model is characterized by AIC value. The results show that LRR, LASSO and PCLR surpass MLE in overcoming partial and full multicollinearity in binary logistic model. PCLR exceeds LRR and LASSO when full multicollinearity occurs in binary logistic model but LASSO and LRR are better used when partial multicollinearity exists in the model.
