Title: Change point detection in partial correlation networks
Authors: Jessica Wai Yin Leung - University of Sydney (Australia) [presenting]
Dmytro Matsypura - University of Sydney (Australia)
Abstract: The focus is on the problem of change point detection in the structure of similarity network over time. The correlation matrix is commonly used in constructing similarity networks. Yet, it is well known that it suffers from the illusion of spurious relationships within the system. Therefore, we adopt the partial correlation matrix as an alternative similarity measure and develop a statistical testing procedure that is tailored to such construct with minimal distributional assumptions. Simulation results show that our test has high statistical power across different sample sizes and dimensions.