A0295
Title: Detecting change points for a multivariate functional data sequence
Authors: Yu-Ting Chen - National Chengchi University (Taiwan) [presenting]
Jeng-Min Chiou - Academia Sinica (Taiwan)
Abstract: Functional data with a multivariate outcome is a common data type. We introduce a procedure for detecting change points for a multivariate functional data sequence based on the developed optimality criterion in defining the functional changes. The approach first searches for change point candidates using the dynamic segmentation that recursively adjusts the endpoints of the subsequences via the optimality criterion. Then, it verifies the statistical significance of these change point candidates by a resampling technique that requires very mild assumptions. We show the consistency property of the algorithm, illustrate the method by a traffic data application, and examine its practical performance through a simulation study.
