| Summary: | 碩士 === 國立成功大學 === 統計學研究所 === 81 === Serveral authors have constructed nonparametric Bayes
estimators for a cumulative distribution function or a
cumulative hazard function based on (possibly right censoring)
data. The prior distributions have, for example, been Dirichlet
processes, beta processes or, more generally, processes neutral
to the right. How robust is the Bayes estimator with respect to
the prior distribution ? Since the parameters of prior
distribution play an important role in Bayesian analysis, in
this article we will use " Monte Carlo Simulation " to make a
comparison on nonparametric empirical Bayes estimators of the
distribution function and cumulative hazard rate under a
Dirichlet process or a beta process prior.
|
|---|
