Department of Statistics Event Calendar

View Full Calendar

Statistics Seminar - Zifeng Zhao (Notre Dame)

Event Type
Seminar/Symposium
Sponsor
Department of Statistics
Virtual
wifi event
Date
Sep 17, 2020   3:30 pm  
Views
89

Abstract: This paper develops a unified, accurate and computationally efficient method for change-point inference in non-stationary spatio-temporal processes. By modeling a non-stationary spatio-temporal process as a piecewise stationary process, we consider simultaneous estimation of the number and locations of change-points, and model parameters in each segment. A composite likelihood-based criterion is developed for change-point and parameters estimation. Asymptotic theories including consistency and distribution of the estimators are derived under mild conditions. In contrast to classical results for fixed-dimension time series that the asymptotic error of the change-point estimator is Op(1), exact recovery of true change-points is possible in the spatio-temporal setting under certain conditions. More surprisingly, the consistency of change-point estimation can be achieved without any penalty term in the criterion function. A computational efficient pruned dynamic programming algorithm is developed for the challenging criterion optimization problem. Simulation studies and an application to U.S. precipitation data are provided to demonstrate the effectiveness and practicality of the proposed method.


Join Zoom Meeting

https://illinois.zoom.us/j/98022381976?pwd=SDBUVTAwTCs5aUhCMVhtdjAzVlpidz09

Meeting ID: 980 2238 1976

Password: 448441

link for robots only