Abstract
This paper examines the Cox proportional hazards model (CPHM) in the presence of multicollinearity. Typically, the maximum partial likelihood estimator (MPLE) is employed to estimate the model coefficients, which works well when the covariates are uncorrelated. However, in various scenarios, covariates are correlated, leading to unstable coefficient estimates with the MPLE. To address this challenge, Liu and ridge estimators have been introduced in the CPHMs. In this paper, we present the Kibria-Lukman estimator as an advancement over existing alternatives and explore its properties. We evaluate the performance of the proposed estimator through Monte Carlo simulations, utilizing mean squared error and mean absolute error as criteria for comparison. Additionally, we demonstrate our proposal advantages through analyzing a medical dataset.