A0844
Title: Multiple testing of local extrema for detection of structural breaks in linear models
Authors: Zhibing He - Yale University (United States)
Dan Cheng - Arizona State University (United States) [presenting]
Yunpeng Zhao - Colorado State University (United States)
Abstract: A new approach to detect structural breaks (change points) based on differential smoothing and multiple testing is presented for long data sequences modeled as piecewise linear functions plus stationary Gaussian noise. As an application of the STEM algorithm for peak detection, the method detects change points as significant local maxima and minima after smoothing and differentiating the observed sequence. The algorithm, combined with the Benjamini-Hochberg procedure for thresholding p-values, provides asymptotic strong control of the False Discovery Rate (FDR) and power consistency, as the length of the sequence and the size of the jumps or slope changes get large. Simulations show that FDR levels are maintained in non-asymptotic conditions.
