Sequential Common Change Detection and Isolation of Changed Panels in Panel Data

Quick detection of common changes is critical in sequential monitoring of multi-stream data where a common change is referred as a change that only occurs in a portion of panels. After briefly reviewing the CUSUM and Shiryayev-Roberts (SR) procedures for a single sequence under an exponential family model, we propose a combined CUSUM-SR procedure that is locally optimal in terms of the delay detection time. The design based on the Average In-control Run Length and comparisons with other procedures are discussed. After a common change is detected, a classifier formed by the post-change parameter estimations is used to isolate the possible candidates for the changed panels, that are then used to estimate the common change-point. To reduce the false discovery rate (FDR), supplementary runs are proposed. Dow Jones 30 Industrial Stock Prices are used for demonstration.

[1]  Yanhong Wu,et al.  Estimation of common change point and isolation of changed panels after sequential detection , 2019 .

[2]  Yanhong Wu.  Supplementary score test for sparse signals in large-scale truncated one-sided sequential tests , 2018 .

[3]  Jay Bartroff,et al.  Multiple Hypothesis Tests Controlling Generalized Error Rates for Sequential Data , 2014, 1406.5933.

[4]  Yanhong Wu.  Inference for post-change mean by a CUSUM procedure , 2006 .

[5]  Yanhong Wu,et al.  A combined SR-CUSUM procedure for detecting common changes in panel data , 2018, Communications in Statistics - Theory and Methods.

[6]  Yanhong Wu.  Inference for Change Point and Post Change Means After a CUSUM Test , 2005 .

[7]  Y. Mei.  Efficient scalable schemes for monitoring a large number of data streams , 2010 .

[8]  D. Siegmund,et al.  A diffusion process and its applications to detecting a change in the drift of Brownian motion , 1984 .

[9]  M. Pollak.  Average Run Lengths of an Optimal Method of Detecting a Change in Distribution. , 1987 .

[10]  Yanhong Wu.  Bias of estimator of change point detected by a CUSUM procedure , 2004 .

[11]  V. Veeravalli,et al.  Asymptotically Optimal Quickest Change Detection in Distributed Sensor Systems , 2008 .

[12]  Georgios Fellouris,et al.  Sequential multiple testing with generalized error control: An asymptotic optimality theory , 2016, The Annals of Statistics.

[13]  Yanhong Wu.  Discussion on “Sequential Design and Estimation in Heteroscedastic Nonparametric Regression” by Sam Efromovich , 2007 .

[14]  Hock Peng Chan,et al.  Optimal sequential detection in multi-stream data , 2015, 1506.08504.

[15]  Yanhong Wu.  False Alarms and Sparse Change Segment Detection by Using a CUSUM Procedure , 2007 .

[16]  D. Siegmund,et al.  Sequential multi-sensor change-point detection , 2012, 2013 Information Theory and Applications Workshop (ITA).

[17]  Aleksey S. Polunchenko.  On the quasi-stationary distribution of the Shiryaev–Roberts diffusion , 2016, 1606.06658.

[18]  J. Bartroff,et al.  Sequential Tests of Multiple Hypotheses Controlling Type I and II Familywise Error Rates. , 2013, Journal of statistical planning and inference.

[19]  Michael Baron,et al.  Sequential tests controlling generalized familywise error rates , 2015 .

[20]  Yanhong Wu,et al.  SUPPLEMENTARY SCORE TEST IN MIXTURE MODEL , 2002 .

[21]  Moshe Pollak,et al.  Asymptotic Exponentiality of the Distribution of First Exit Times for a Class of Markov Processes with Applications to Quickest Change Detection , 2006 .