Bayesian change-point modeling with segmented ARMA model

Time series segmentation is a dynamic sphere of research that has attracted attention in diverse topics. The aim of this thesis is to segment time series data using a Bayesian approach and develop methods to segment multiple parallel time series data. This thesis used a novel methodology called the Bayesian change-point segmented ARMA model, which estimates the change-point locations by segmenting a time series using an ARMA model. It also uses a highly efficient technique (Generalized Gibbs Sampler) to generate samples from a posterior distribution. Three alternative generalizations of the Bayesian change-point segmented ARMA model were developed to segment multiple parallel time series data.