Generalization of Three Weibull Extensions with Applications in Renal Failure Data

• Activity Number: 299 - Survival and Recurrent Events in Epidemiology
• Type: Contributed
• Date/Time: Tuesday, August 1, 2017 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Nonhle Channon Mdziniso* and Kahadawala Cooray
• Companies: Central Michigan University and Central Michigan University
• Keywords: coverage probability ; goodness-of-fit ; Kaplan-Meier curve ; maximum likelihood estimation ; censored data

Abstract

The three parameter Weibull extensions: generalized gamma (GG), exponentiated Weibull (EW), and Odd Weibull (OW) distributions, which are capable of modeling data that exhibit five major hazard shapes are compared using Vuong and empirical distribution function (EDF) tests. Our results show that the OW is different from GG and EW distributions. Therefore, we extend both GG and EW distributions with an extra parameter in order to have the OW distribution as a submodel. The two new four-parameter extensions, gamma Odd Weibull (GOW) and exponentiated Odd Weibull (EOW) distributions produce more advanced hazard shapes: M-, W-, S-, and inverse-S shapes, than the hazard shapes arising from either GG or EW distributions. To illustrate hazard flexibility of the EOW and GOW distributions, an example on renal failure data is analyzed.