Authors:
Dirk Pauli
1
;
Jens Feller
1
;
Bernhard Mauersberg
1
and
Ingo J. Timm
2
Affiliations:
1
FCE Frankfurt Consulting Engineers GmbH, Germany
;
2
University of Trier, Germany
Keyword(s):
Change-point detection, Hypothesis testing, Chernoff bounds, Binomial distribution, Additive changes, Nonadditive changes, Multiple structural break detection.
Related
Ontology
Subjects/Areas/Topics:
Adaptive Signal Processing and Control
;
Business Analytics
;
Change Detection
;
Data Engineering
;
Informatics in Control, Automation and Robotics
;
Signal Processing, Sensors, Systems Modeling and Control
Abstract:
In this paper a new method for detecting multiple structural breaks, i.e. undesired changes of signal behavior, is presented and applied to real-world data. It will be shown how Chernoff Bounds can be used for highperformance hypothesis testing after preprocessing arbitrary time series to binary random variables using k-means-clustering. Theoretical results from part one of this paper have been applied to real-world time series from a pharmaceutical wholesaler and show striking improvement in terms of forecast error reduction, thereby greatly improving forecast quality. In order to test the effect of structural break detection on forecast quality, state of the art forecast algorithms have been applied to time series with and without previous application of structural break detection methods.