Bias Reduction of Maximum Likelihood Estimation In Generalized Ramos-Louzada Distribution

The generalized Ramos-Louzada distribution (GRL) is a potential option for modeling the survival data which considers flexibility while modeling data with decreasing, increasing, reversed-J shaped, and J-shaped hazard rate functions. It should however be noted that the most widely used technique of parameter estimation of GLR is the maximum likelihood technique (MLE). While the MLE is quite efficient in large samples, they are known to be highly biased for small samples. We are therefore forced to come up with essentially nonbiased estimators for GRL parameters. Particularly, we investigate two procedures of bias correction: bootstrap and analytical methods in order to minimize MLE bias up to the second-degree. Two real-world data applications and Monte Carlo simulations are used to countercheck the conclusions of these methods. Simulation results highlight the proposed approaches’ performance, which is significantly less biased than the MLE, which is highly Biased. In small sample sizes, the bias is eliminated to about the half of the ori