A modified ridge estimator in Cox regression model

The most used approach for survival data is the Cox proportional hazards regression model. Multicollinearity, however, is known to have a detrimental impact on the variance of maximum likelihood estimator of the Cox regression coefficients. It has been repeatedly shown that the ridge estimator is a desirable shrinking technique to lessen the consequences of multicollinearity. In this paper, a linearized ridge estimator is developed to transform the biasing parameter of the ordinary ridge estimator to a linearized version and study its performance in the Cox regression model under multicollinearity. Our Monte Carlo simulation findings and the real data application indicate that the suggested estimator may significantly reduce mean squared error in comparison to other competing estimators.