摘要
This paper introduces a new rich family of distributions based on mixtures and the so-called Marshall-Olkin family of distributions.It includes a wide variety of well-established mixture distributions,ensuring a high ability for data fitting.Some distributional properties are derived for the general family.The Weibull distribution is then considered as the base-line,exhibiting a pliant four-parameter lifetime distribution.Five estimation methods for the related parameters are discussed.Bootstrap confidence intervals are also considered for these parameters.The distribution is reparametrized with location-scale parameters and it is used for a lifetime regression analysis.An extensive simulation is carried out on the esti-mation methods for distribution parameters and regression model parameters.Applications are given to two practical data sets to illustrate the applicability of the new family.
