Reparameterized Flexible Weibull Distribution with Some Applications

Abstract

A reparameterized flexible Weibull distribution indexed by median and a shape parameter is proposed for the development of regression models which includes the possibility of censored data. The reparameterization permits a straightforward interpretation of the regression coefficients in terms of the median. Model estimation is implemented via Classical and Bayesian approaches, and Monte Carlo simulations are carried out in order to evaluate the estimators performances for finite samples. In addition, the model is applied to three real data sets.

KEY WORDS AND PHRASES:

• Bayesian estimation
• censored data
• flexible Weibull
• maximum likelihood
• survival analysis