Bayesian estimation of change points using the general linear model

Abstract

Bayesian analysis is applied to the general linear model to develop a framework for studying different types of change in the mean value of time series and linear regressions. The output of the Bayesian analysis is the posterior distribution of change point location and amplitude. This information provides a rational and relatively objective basis for making decisions as to where to locate a change point. Several examples of hydrological applications are presented to demonstrate the utility of the methodology.