A novel framework for detecting multiple change points in functional data sequences (2505.15188v2)

Abstract: Detecting multiple change points in functional data sequences has been increasingly popular and critical in various scientific fields. In this article, we propose a novel two-stage framework for detecting multiple change points in functional data sequences, named as detection by Group Selection and Partial F-test (GS-PF). The detection problem is firstly transformed into a high-dimensional sparse estimation problem via functional basis expansion, and the penalized group selection is applied to estimate the number and locations of candidate change points in the first stage. To further circumvent the issue of overestimating the true number of change points in practice, a partial F-test is applied in the second stage to filter redundant change points so that the false discovery rate of the F-test for multiple change points is controlled. Additionally, in order to reduce complexity of the proposed GS-PF method, a link parameter is adopted to generate candidate sets of potential change points, which greatly reduces the number of detected change points and improves the efficiency. Asymptotic results are established and validated to guarantee detection consistency of the proposed GS-PF method, and its performance is evaluated through intensive simulations and real data analysis, compared with the state-of-the-art detecting methods. Our findings indicate that the proposed GS-PF method exhibits detection consistency in different scenarios, which endows our method with the capability for efficient and robust detection of multiple change points in functional data sequences.