Bayes-type tests for constancy of autoregressive parameters

✍ Scribed by Anna M. Lurie; Nagaraj K. Neerchal

• Publisher: John Wiley and Sons
• Year: 1999
• Tongue: English
• Weight: 256 KB
• Volume: 10
• Category: Article
• ISSN: 1180-4009
• DOI: 10.1002/(sici)1099-095x(199911/12)10:6<737::aid-env386>3.0.co;2-g

✦ Synopsis

Environmental data collected over time can often be described by models with parameters changing suddenly at some unknown point in time. In statistical literature, problems concerning the estimation and testing for the validity of such models are collectively referred to as `change-point problems'. In this paper we focus on the change-point problems occurring in autocorrelated data. Bayes-type tests are derived for the problem of testing for parameter changes at an unknown time in a general autoregressive process. Test statistics and their asymptotic distributions are obtained. In a simulation study, it is indicated that Bayestype tests have better power than likelihood ratio tests for detecting small changes. Methodology is illustrated using ozone concentration data.