Greedy Segmentation for a Functional Data Sequence

We present a new approach known as greedy segmentation (GS) to identify multiple changepoints for functional data sequence. The proposed changepoint detection criterion links detectability with the projection onto suitably chosen subspace and locations. estimator identifies true any predetermined number of candidates, either over-reporting or under-reporting. This theoretical finding supports GS estimator, which can be efficiently obtained in manner. estimator’s consistency holds without being restricted conventional at most one condition, it is robust relative positions changepoints. Based on test statistic’s asymptotic distribution leads novel algorithm, locations Using intensive simulation studies, we compare finite sample performance other competing methods. also apply our method temporal weather datasets.