Modelo exponencial por partes via modelos partição produto

• Ano de defesa: 2006
• Autor(a) principal: Fabio Nogueira Demarqui
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: Exponencial; Partes; Estatística; Teoria bayesiana de decisão estatistica; Teoria da estimativa
• Link de acesso: http://hdl.handle.net/1843/RFFO-7HQRJY

One of the greatest difficulty related to the use of the piecewise exponential model is to find the grid of time-points needed in its definition. In general, the number of intervals in such grid and the position of their endpoints are ad-hoc choices. We introduce a full Bayesian approach for the piecewise exponential distribution in which the endpoints and the number of intervals are random variables. We estimate the failure rate using the proposed procedure and compare the results with the piecewise exponential estimates. A sensitivity analysis for the proposed model is provided considering different prior cohesions and different prior distributions for the failurerate. We apply the methodology to analyse two real data set.