Modelos para dados de sobrevivência na presença de diferentes esquemas de ativação baseados na distribuição geométrica
In this thesis new families of survival distributions are proposed. Those distributions are derived by assuming a latent activation structure to explain the occurrence of the event of interest. In general, the competitive causes may have different activation mechanisms. Here we assume three different ones, namely, fisrt, random and last actvation mechanisms. The presence of cure fraction are also addressed in two contexts. The models assumed that the number of causes follows a Geometric distribution and the lifetime for these causes follows an Exponential distribution, and a Gamma Generalized distribution. The properties of the proposed distributions are discussed, including a formal proof of its probability density function and explicit algebraic formulas for its reliability and failure rate functions, moments, order statistics and modal value. Inferetial procedure is based on frequentist and Bayesian perspectives. Moreover, Bayesian case influence diagnostics based in -divergence, with include Kulback Leibler divergence measure as a particular case, are developed. Simulation studies are performed and experimental results are illustrated based in real datasets.