CMStatistics 2021: Start Registration
View Submission - CMStatistics
B0735
Title: Spherical autoregressive change-point detection with applications Authors:  Federica Spoto - Sapienza University of Rome (Italy) [presenting]
Alessia Caponera - EPFL (Switzerland)
Pierpaolo Brutti - University of Rome - Sapienza (Italy)
Abstract: Spatio-temporal processes arise very naturally in a number of different applied fields, like Cosmology, Astrophysics, Geophysics, Climate and Atmospheric Science. In most of these areas, the detection of structural breaks or regime shifts in the data stream is key. To this end, in the present work, we aim at generalizing the recently introduced SPHAR(p) process by allowing for temporal changes in its functional parameters and variability structure. Our approach, which intrinsically integrates the spatial and temporal dimensions, could give multiscale insights into both the global and local behavior of changes, and its performance will be tested on a real dataset of global surface temperature anomalies.