A new extended log-Weibull regression: Simulations and applications

Authors : Gauss M CORDEİRO, Mh TAHİR, Julio Cezar SOUZA VASCONCELOS, Edwin Mm ORTEGA, M Adnan HUSSAİN

Pages : 855-871

Doi:10.15672/hujms.779454

View : 26 | Download : 6

Publication Date : 2021-06-07

Article Type : Research Paper

Abstract :The induction of one or more parameterinsert ignore into journalissuearticles values(s); in parent distributions opened new doors for flexible modeling in modern distribution theory. Among well-established generalized insert ignore into journalissuearticles values(G); classes for flexible modeling, the exponentiated-G, Marshall-Olkin-G and odd log-logistic-G families offer induction of one additional parameter while the beta-G and Kumaraswamy-G classes offer two extra shape parameters. The Marshall-Olkin-odd-loglogistic-G insert ignore into journalissuearticles values(MOOLL-G); family serves as an alternative to the beta-G and Kumaraswamy-G classes. A new motivation for the MOOLL-G family for competing risk scenarios, some useful properties, and parameter estimation are addressed. The new log-MOOLL-Weibull regression is useful for analysis of real life data. The accuracy of the estimates and the residuals is addressed via Monte Carlo simulations. The presented models outperform some other well-known models.
Keywords : generalized family, Marshall Olkin method, maximum likelihood, regression model, Weibull distribution