Modellbaserad prediktion av säkerhetspolitiska händelser

Authors:

  • Jonas Clausen Mork
  • Carina Gunnarson
  • Michael Jonsson
  • Farzad Kamrani

Publish date: 2020-06-11

Report number: FOI-R--4954--SE

Pages: 85

Written in: Swedish

Research areas:

  • Säkerhetspolitik

Keywords:

  • prediction
  • security policy
  • statistics
  • machine learning

Abstract

This report studies four model-based conflict prediction systems - ICEWS, PVRAM, CRAM and ViEWS - with regard to overall architecture, focus and technology choices. FOI's production in the area of African security policy studies has also been examined. These two studies provide the basis for a discussion concerning the possibility, and utility, of FOI creating a capability of its own regarding the construction and use of predictive systems in the domain of security policy. The study concludes, among other things, that technological progress in several areas have markedly improved the possibilities to construct capable predictive systems focusing on security policy-related events. There is also a growing body of experience from constructing these kinds of systems within the academic sphere, as well as in private companies. Judging from the focus and content of FOI's publications concerning African security policy, prediction is a recurrent and important theme. The assessment is therefore that the agency's security policy studies would benefit from a strengthened predictive capability. Finally, the prospects at FOI for construction and management of predictive systems in the domain of security policy look favorable, even if their development is likely to entail considerable challenges.

Share page on social media