摘要：The ve parameter Kumaraswamy generalized gamma model (Pascoaet al., 2011) includes some important distributions as special cases andit is very useful for modeling lifetime data. We propose an extended versionof this distribution by assuming that a shape parameter can take negativevalues. The new distribution can accommodate increasing, decreasing, bathtuband unimodal shaped hazard functions. A second advantage is that italso includes as special models reciprocal distributions such as the reciprocalgamma and reciprocal Weibull distributions. A third advantage is thatit can represent the error distribution for the log-Kumaraswamy generalizedgamma regression model. We provide a mathematical treatment ofthe new distribution including explicit expressions for moments, generatingfunction, mean deviations and order statistics. We obtain the moments ofthe log-transformed distribution. The new regression model can be usedmore e ectively in the analysis of survival data since it includes as submodelsseveral widely-known regression models. The method of maximumlikelihood and a Bayesian procedure are used for estimating the model parametersfor censored data. Overall, the new regression model is very usefulto the analysis of real data.
