Survival Analysis Model And Its Application In Reliability Evaluation Using WINBUGS

Survival analysis is one of the most significant advancements of mathematical statistics in the last quarter of the20th century. Although reliability was conceived as a major application field by the mathematicians who pioneered survival analysis, survival analysis failed to establish itself as a major tool for reliability analysis in a long time. In fact, comparing the major mathematical models of both fields, survival analysis and reliability theory essentially address the same mathematical problems. Therefore, survival analysis should become a major mathematical tool for reliability analysis and related fields.In this paper, we employ the survival analysis models such as parametrical models and semi-parametrical models, propose the estimation and comparison methods based on the Bayesian theory and MCMC sampling algorithm with WINBUGS software. We give the details of the Doodle modeling processes with the censored data. Then apply the exponential regression model, the Weibull regression model and the log normal regression model as well as the COX regression model and AFT model into the reliability evaluation with the real data. The results are compared and discussed.