Data Science Seminar: Detecting Local and Global Changes in Terrorism Incidence and the Effects of the COVID-19 Pandemic11 Jun 2021, by Sponsored events in
3 December 2020 | 11:00 (Online)
By: Sam Tickle, Heilbronn Fellow, University of Bristol
Over the past fifty years, terrorism has become highly globalised. Where before there was a collection of highly disparate groups with local or national grievances, the presence of terror groups with a more ubiquitous presence is now an unfortunate assumption of necessity for today’s policymakers. One quantitative means of experiencing this story is through the Global Terrorism Database (GTD). The GTD is an open-source collation of terrorist events which have occurred worldwide since 1970. A natural question which arises from this very rich source of information concerns the presence of changepoints: namely, are there specific points or periods of time in the recent past in which the probability of a terrorist attack has increased, either in a specific global region or worldwide?
This requirement to distinguish between local and global changes motivates the use of SUBSET, a new method for detecting changes in high-dimensional datasets. I’ll talk about a few of the properties of SUBSET (in particular, its ability to detect and distinguish between localised and global changes); discuss the results of its application to the GTD; and conclude with a second application involving an analysis of the effects of the COVID-19 in various European countries.