The ViEWS2018 model

Overview of the methodology and data infrastructure behind the legacy conflict prediction model ViEWS2018
ViEWS2018 was the first operational conflict prediction model employed by the VIEWS system. The pilot model was in use between 2018-2020. 
Example prediction plot from the ViEWS2018 model
Example prediction plot from the ViEWS2018 model. Predicted probability of at least one observed fatality from state-based conflict in June 2018, based on input data up to April 2018.
  • This is a legacy model
Please note that ViEWS2018 is a legacy model. The model codebase remains accessible via the open-source GitHub repository OpenViEWS for transparency, but neither the model nor its underlying data infrastructure is supported or maintained. Please consult the operational model documentation for up-to-date information about the live prediction system. 

OVERVIEW OF THE PREDICTION MODEL

The ViEWS2018 model

  • Hover over the boxes for more information.

UPDATE SCHEDULE

Monthly

PREDICTED TYPE(S) OF VIOLENCE

State-based conflict, non-state conflict, and one-sided violence

FORECASTING WINDOW

1-36 months ahead

PREDICTED OUTCOMES

Probability of exceeding pre-determined conflict thresholds

COUNTRY-LEVEL COVERAGE

Africa

SUB-NATIONAL COVERAGE

Africa

Model documentation

ViEWS: A political violence early-warning system

Håvard Hegre, Marie Allansson, Matthias Basedau, Michael Colaresi, Mihai Croicu, Hanne Fjelde, Frederick Hoyles, Lisa Hultman, Stina Högbladh, Remco Jansen, Naima Mouhleb, Sayyed Auwn Muhammad, Desirée Nilsson, Håvard Mokleiv Nygård, Gudlaug Olafsdottir, Kristina Petrova, David Randahl, Espen Geelmuyden Rød, Gerald Schneider, Nina von Uexkull, and Jonas Vestby.
Year: 2019
Journal of Peace Research 56(2), 155-174

Data infrastructure

The ViEWS1 data infrastructure supporting the ViEWS2018 model

The ViEWS2018 model was built on and supported by the ViEWS1 data infrastructure. It is documented alongside the model codebase in the OpenViEWS GitHub repository. Please note that the repository remains open for transparency, but is no longer supported or maintained.