Title: On the support of some random contiguous set partitions for change point analysis
Authors: Jose Javier Quinlan Binelli - Pontificia Universidad Catolica de Chile (Chile) [presenting]
Alejandro Jara - Pontificia Universidad Católica de Chile (Chile)
Mauricio Castro - Pontificia Universidad Catolica de Chile (Chile)
Abstract: In Bayesian change point analysis for univariate time series, prior distributions on the space of contiguous set partitions play a key role for change point detection. In this context, mixtures of such distributions are appealing candidates in terms of diversity in the specification of prior beliefs. However, how flexible are these prior processes to describe the number and locations of possible change-points? We will address the previous question in terms of weak support.