Title: Multiple changepoints detection for a functional data sequence
Authors: Yu-Ting Chen - National Chengchi University (Taiwan) [presenting]
Jeng-Min Chiou - Academia Sinica (Taiwan)
Abstract: To detect multiple changes in a sequence of functional data, an algorithm called simultaneously dynamic segmentation (SDS) is proposed which performs dynamic segmentation simultaneously concerning different initial segments. SDS is free from the at-most-one-changepoint assumption and searches for the global minimum of the derived criterion. We also derive the null distribution of the objective function to determine the number of changepoints either forwardly or backwardly. We demonstrate the flexibility and validity of SDS through a simulation study and apply SDS to a traffic data set to find locations of changes in traffic conditions.