Performance of bayesian using conjugate prior estimator for Weibull right censored survival data

Abstract

Background and Objective: The Weibull distribution is widely used to model and to analyze data on the survival time. Bayesian estimation approach has received much attention as it has been in contention with other estimation methods. In this study, it was examined the performance of the Bayesian estimator using conjugate prior information for estimating the parameters of Weibull distribution with censored survival data for dengue fever (DF). Materials and Methods: Through the simulated Weibull distributed survival dataset, the performance of conjugate estimator for estimating the Weibull distribution parameters can be checked before applying it to the DF survival dataset in Makassar, Indonesia. Statistical analysis of the simulation data and collected DF data were analyzed through summary tables and Markov Chain Monte Carlo method via Gibbs sampling algorithm. It was performed using R version 3.3.3 and Win BUGS. Results: Based on the simulation study, the mean of posterior means of all Weibull distribution parameter estimates are still reasonably accurate. After fitting the Weibull models to the DF survival times dataset using the conjugate prior distribution, the age factor substantially described DF patients survival times and had a positive effect on the estimated survival time. Conclusion: To choose sample size and censoring level, the estimates generated by the conjugate prior do not only depend on the data but also on the parameters of the prior distribution. The amount of uncensored data must be more than one in order to obtain an estimate of greater than zero. The results of estimated parameters of Weibull model using conjugate prior either with simulated survival data or the DF data is good.

Access to Document

Other files and links

• Link to publication in Scopus
• Open Access Version Available

Fingerprint