13/11/2020 - Seminário - A Multipartition model for multiple change-point identification - Palestrante: Rosangela Loschi (UFMG)

Seminário Conjunto UFSCar/USP

Data e horário:
13/11/2020 às 14h

Local:
Google Meet - meet.google.com/xok-emqv-imd 

Título:
A Multipartition model for multiple change-point identification 

Palestrante:
Rosangela Loschi (UFMG)

Resumo: 

The main goals in multiple change point problems are the estimation of the number and positions of the change points as well as the regime structure in the clusters induced by the change points. The product partition model (PPM)  is a widely used approach to detect multiple change points. The PPM assumes that the change points split the set of time points in random instants defining a partition of the time axis. As changes in different parameters may occur in  different times, the PPM fails in identifying the parameters that experienced the changes. A similar problem may occur when detecting changes in multivariate time series. We introduce a multipartition model to detect multiple change points occurring in several parameters. The proposed model assumes that the change points experienced by each parameter generate a different random partition of the time axis, which enables to identify which and when the parameters change.   We propose a partially collapsed Gibbs sampler scheme to generate from the posterior distributions. We apply the proposed model to identify multiple change points in Normal means and variances. We evaluate the performance of the proposed model through a Monte Carlo simulation and considering real data applications. Its performance is compared with that of previously proposed models. These studies show that the proposed model is competitive and enriches the analysis of change point problems.