Bayesian comparative study on binary time series. Issue 14 (22nd September 2018)

• Record Type: Journal Article
• Authors: Paul, Erina; Maity, Arnab Kumar; Maiti, Raju
• Is Part Of: Journal of statistical computation and simulation. Volume 88:Issue 14(2018)
• Page Start: 2811
• Page End: 2826
• Publication Date: 2018-09-22
• Subjects: DIC -- Jeffreys prior -- log-marginal likelihood -- misclassification error rate -- normal prior -- prediction -- Student's t prior; 62F15; Mathematical statistics -- Data processing -- Periodicals; Digital computer simulation -- Periodicals; 519.5028505
• DOI: 10.1080/00949655.2018.1488256
• Languages: English
• ISSNs: 0094-9655

Abstract

ABSTRACT: In this paper, we consider the Bayesian analysis of binary time series with different priors, namely normal, Students' t, and Jeffreys prior, and compare the results with the frequentist methods through some simulation experiments and one real data on daily rainfall in inches at Mount Washington, NH. Among Bayesian methods, our results show that the Jeffreys prior perform better in most of the situations for both the simulation and the rainfall data. Furthermore, among weakly informative priors considered, Student's t prior with 7 degrees of freedom fits the data most adequately.