Seol mar théacs é seo: A Double-Penalized Estimator to Combat Separation and Multicollinearity in Logistic Regression