A0973
Title: Change point detection for random objects using distance profiles
Authors: Paromita Dubey - University of Southern California (United States)
Minxing Zheng - University of Southern California (United States)
Paromita Dubey - University of Southern California (United States) [presenting]
Abstract: The purpose is to introduce a new powerful scan statistic and an associated test for detecting the presence and pinpointing the location of a change point within the distribution of a data sequence where the data elements take values in a general separable metric space. These change points mark abrupt shifts in the distribution of the data sequence. The method hinges on distance profiles, where the distance profile of an element is the distribution of distances from it as dictated by the data. The approach is fully non-parametric and universally applicable to diverse data types, including distributional and network data, as long as distances between the data objects are available. Through comprehensive simulation studies encompassing multivariate data, bivariate distributional data, and sequences of graph Laplacians, the effectiveness of the approach is demonstrated in both change point detection power and estimating the location of the change point. The method is applied to real datasets, including U.S. electricity generation compositions and Bluetooth proximity networks, underscoring its practical relevance.
