Dealing With Non-Stationarities in Violence Data Using Empirical Mode Decomposition

16-2-2.jpg
16-2-2.jpg

Dealing With Non-Stationarities in Violence Data Using Empirical Mode Decomposition

9.95

Author(s): Peter Dobias; James A. Wanliss
No pages: 4
Year: 2013
Article ID: 16-2-2
Keywords: EMD, empirical mode decomposition, non-stationarities, training and analysis
Format: Electronic (PDF)

Add To Cart

Abstract: Empirical mode decomposition (EMD) has been used across a variety of different fields such as biology and plasma physics. This paper presents the application of EMD to time series of security incidents from Afghanistan. This methodology enables separation of different modes intrinsic to the data, thus enabling better understanding of various overlying trends when compared to more conventional methodologies, and it provides a venue for the in-depth analysis of multi-scale dynamical properties of the data. In particular, the approach does not require a priori assumptions about time dependence of various data sub-components, such as periodicity of variations, thus providing a superior approach to conventional methods of analyzing violence data such as seasonal decomposition not only for the analysis of stochastic and fractal properties of the data, but even for a more conventional analysis of violence trends in direct support of military operations.