Summary reports of the monthly conflict forecasts
Browse all peer-reviewed articles, reports, and other publications from the VIEWS team
Download the VIEWS forecasts, corresponding input- and replication data, and consult our GIS resources
Explore the codebase behind the VIEWS models and their data infrastructure
Press photos and logos

Publications 

Peer-reviewed articles
The ‘conflict trap’ reduces economic growth in the shared socioeconomic pathways  
Kristina Petrova, Gudlaug Olafsdottir, Håvard Hegre, and Elisabeth A Gilmore
Environmental Research Letters, 2023
Employing local peacekeeping data to forecast changes in violence 
Lisa Hultman, Maxine Leis & Desirée Nilsson 
International Interactions 48(4), 2022

Predicting Escalating and De-Escalating Violence in Africa Using Markov Models
David Randahl & Johan Vegelius
International Interactions 48(4), 2022
United They Stand: Findings from an Escalation Prediction Competition
Paola Vesco, Håvard Hegre, Michael Colaresi, Remco Bastiaan Jansen, Adeline Lo, Gregor Reisch & Nils B. Weidmann
International Interactions 48(4), 2022
Lessons from an Escalation Prediction Competition
Håvard Hegre, Paola Vesco, and Michael Colaresi
International Interactions 48(4), 2022
Climate variability, crop and conflict: Exploring the impacts of spatial concentration in agricultural production
Paola Vesco, Matija Kovacic, Malcolm Mistry, Mihai Croicu
Journal of Peace Research, 2021
ViEWS2020: Revising and evaluating the ViEWS political Violence Early-Warning System
Hegre, Håvard, Curtis Bell, Michael Colaresi, Mihai Croicu, Frederick Hoyles, Remco Jansen, Angelica Lindqvist-McGowan, David Randahl, Espen Geelmuyden Rød, Maxine Ria Leis and Paola Vesco,
Journal of Peace Research
58(3), 2021
Can We Predict Armed Conflict? How the First 9 Years of Published Forecasts Stand Up to Reality
Håvard Hegre, Håvard Mokleiv Nygård, Peder Landsverk
International Studies Quarterly, 2021
ViEWS: A political Violence Early Warning System
Hegre, Håvard, Marie Allansson, Matthias Basedau, Michael Colaresi, Mihai Croicu, Hanne Fjelde, Frederick Hoyles, Lisa Hultman, Stina Högbladh, Remco Jansen, Naima Mouhleb, Sayeed Auwn Muhammad, Desirée Nilsson, Håvard Mokleiv Nygård, Gudlaug Olafsdottir, Kristina Petrova, David Randahl, Espen Geelmuyden Rød, Gerald Schneider, Nina von Uexkull, and Jonas Vestby.
Journal of Peace Research
56(2), pp. 155-174, 2019
Evaluating the conflict-reducing effect of UN peacekeeping operations
Hegre, Håvard, Lisa Hultman, and Håvard Mokleiv Nygård
Journal of Politics 
81(1), 2019
Forecasting civil conflict along the shared socioeconomic pathways
Håvard Hegre, Halvard Buhaug, Katherine V. Calvin, Jonas Nordkvelle, Stephanie T. Waldhoff, and Elisabeth Gilmore
Environmental Research letters
11(5) 054002, 2016
Working papers

The VIEWS Working Paper Series

Working papers related to one or more of the VIEWS projects – the conflict prediction system ViEWS and the impacts projects Societies at Risk and ANTICIPATE – will from the fall of 2022 onwards be published in a dedicated series on the Uppsala University publishing platform, accessible via the link above.

Previous working papers 

Predicting the impact of armed conflict on vulnerability: a Machine Learning approach
Mariagrazia D’Angeli and Paola Vesco
Paper presented at the German Institute of Development and Sustainability (IDOS) workshop in Bonn, Germany, on 24-25 October 2022
Too hot to handle? Climate shocks, societal vulnerability and conflict forecasting
Vesco, Paola, Håvard Hegre, and Ole Magnus Theisen
Paper presented at the ISA Annual Convention 2022 in Nashville, Tennessee
A Review and Comparison of Conflict Early Warning Systems
Espen Geelmuyden Rød, Tim Gåsste, and Håvard Hegre
Working Paper, Uppsala University, 2022
Predicting Armed Conflict Using Protest Data
Espen Geelmuyden Rød, Håvard Hegre, and Maxine Leis
Working Paper, Uppsala University, 2022
A Climate of War or Peace? The Effect of Droughts on Conflict Dynamics
Paola Vesco
Paper presented at the virtual ISA Annual Convention 2021
Appendix
The ‘conflict trap’ reduces economic growth in the Shared Socioeconomic Pathways
Gilmore, Elisabeth, Håvard Hegre, Gudlaug Olafsdottir, and Kristina Petrova
Working Paper, Uppsala University, 2021
Inference with extremes: Accounting for Extreme Values in Count Regression Models
Randahl, David, and Johan Vegelius
Typescript Uppsala University, 2021
A Fast Spatial Multiple Imputation Procedure for Imprecise Armed Conflict Events
Croicu, Mihai, and Håvard Hegre
Paper presented at the 59th Annual Convention International Studies Association in San Francisco, California, 2018.
Early ViEWS: A disaggregated, open-source violence early-warning system
Colaresi, Michael, Håvard Hegre, and Jonas Nordkvelle
Paper presented at the American Political Science Association annual meeting in Philadelphia, 1 September 2016.
Reports

Policy papers

Forecasting fatalities in armed conflict: Forecasts for April 2022-March 2025 (overview of the fatalities model)
Håvard Hegre, Angelica Lindqvist-McGowan, James Dale, Mihai Croicu, David Randahl & Paola Vesco

Report, Uppsala University, 2022
Related project: ViEWS-UK FCDO
Forecasting armed conflict in the Sahel: Forecasts for November 2021–October 2024
Hegre, H., Angelica Lindqvist-McGowan, Paola Vesco, Remco Jansen, and Malika Rakhmankulova
Report, Uppsala University, 2022
Related project: ViEWS-UNHCR (Sahel Predictive Analytics project)
Understanding the potential linkages between climate change and conflict in the Arab region
Ole Magnus Theisen, Håvard Hegre, Halvard Buhaug, Stefan Döring, Remco Jansen, Angelica Lindqvist-McGowan, Ida Rudolfsen, Paola Vesco, Joaquin Salido Marcos, Yara Acaf, Pattile Nahabedian, and Lubna Ismail
United Nations publication, ESCWA, 2021
Related project: ViEWS-ESCWA

Technical papers (model descriptions)

Forecasting fatalities (technical report on the fatalities model)
Håvard Hegre, Forogh Akbari, Mihai Croicu, James Dale, Tim Gåsste, Remco Jansen, Peder Landsverk, Maxine Leis, Angelica Lindqvist-McGowan, Hannes Mueller, Malika Rakhmankulova, David Randahl, Christopher Rauh, Espen Geelmuyden Rød & Paola Vesco
Report, Uppsala University, 9 June 2022
Related project: ViEWS-UK FCDO

Downloads and data access

Conflict data (actuals)
The VIEWS forecasts are informed by, and evaluated against, UCDP GED and UCDP Candidate Events data on observed conflict from 1989 onwards. The UCDP “best estimates” of battle-related deaths per conflict event and type of violence (state-based conflict, non-state conflict, and one-sided violence) are aggregated to the VIEWS levels of analyses at a monthly resolution, i.e. UCDP conflict observations/actuals per PRIO-GRID-month and country-month. This is sometimes referred to as “observations” or “actuals” of/on the “VIEWS outcomes”. 
Historic conflict data on the VIEWS outcomes, used in a set of academic publications, can be found below. For updated conflict data aggregated to the VIEWS units and levels of analysis, please submit a request to views@pcr.uu.se. 
Over the fall of 2022, monthly updates of conflict data (as well as other input data) will also be made available via the VIEWS API. 
Observed conflict 1989-2018, ViEWS units of analysis (ViEWS outcomes), aggregated to the PRIO-GRID-month level The dataset covers UCDP conflict data on the ViEWS outcomes for January 1989 to December 2018, and has a one month resolution. Data for 2018 originates from “UCDP candidate events”. A description of the data is available in the presentation article of ViEWS in Journal of Peace Research.
Codebook (pdf)
Data (.zip)
Please cite: Hegre, Håvard, Mihai Croicu, Kristine Eck, and Stina Högbladh, 2018. “Introducing the UCDP-Candidate Events Dataset and the ViEWS Outcomes dataset. Monthly updated organized violence data in the form of events data as well as aggregated to the country-month and PRIO-GRID-month level”. Typescript Uppsala University.
Observed conflict 1989-2018, ViEWS units of analysis (ViEWS outcomes), aggregated to the country-month level The dataset covers UCDP conflict data on the ViEWS outcomes for January 1989 to December 2018, and has a one month resolution. Data for 2018 originates from “UCDP candidate events”. A description of the data is available in the  presentation article of ViEWS in Journal of Peace Research.
Codebook (pdf)
Data (.zip) – Currently unavailable online, please contact the ViEWS team for the file.
Please cite: Hegre, Håvard, Mihai Croicu, Kristine Eck, and Stina Högbladh, 2018. “Introducing the UCDP-Candidate Events Dataset and the ViEWS Outcomes dataset. Monthly updated organized violence data in the form of events data as well as aggregated to the country-month and PRIO-GRID-month level”. Typescript Uppsala University.
Imputed data on observed conflict 1989-2018, ViEWS units of analysis (ViEWS outcomes), aggregated to the PRIO-GRID-month level A dataset containing 5 multiple imputations of the UCDP conflict data on the ViEWS outcomes (units of analysis) at the PRIO-GRID level for January 1989 to December 2018, for conflict events that do not resolve to a precise PRIO-GRID-cell. The methodology for imputations is described in Mihai Croicu and Håvard Hegre’s 2018 paper ” A Fast Spatial Multiple Imputation Procedure for Imprecise Armed Conflict Events”.
Codebook (pdf)
Data (.zip)
Please cite: Mihai Croicu and Håvard Hegre (2018) ‘A Fast Spatial Multiple Imputation Procedure for Imprecise Armed Conflict Events’, Paper presented at the 59th Annual Convention International Studies Association, San Francisco, California
Forecasts

RETRIEVE AND REQUEST FORECASTS

The continuous prediction models (2022- )
Ensemble results from the continuous prediction models in use from 2022 onwards – provided in the form of both continuous and dichotomous outcomes – as well as selected results from our interpretative surrogate models, are publicly available via the ViEWS API ( api.viewsforecasting.org). Please consult the  API wiki page to get started, coupled with the  viewsforecasting GitHub repository and corresponding  changelog for more information about the models behind each each data release and the naming convention for these releases. 
Complete results from the continuous prediction models can be accessed via the web-based client viewser, which allows you to communicate directly with the ViEWS3 data infrastructure at your convenience. Such access requires a certificate issued by the ViEWS team. If you or your organisation(s) are interested in using this service, please contact views@pcr.uu.se for advice and assistance. 
The dichotomous prediction models (2017-2021)
A selection of forecasts from the dichotomous prediction model under the now deprecated ViEWS2 infrastructure (OpenViEWS2, in use 2018-2021) are available for download below, through the  interactive forecasting dashboard, and via the ViEWS API (the data releases named r_[year]_[month]_01).
For access to other datasets from these prediction models and data infrastructure, or from the 2017-2018 prediction models under the deprecated ViEWS1 infrastructure ( OpenViEWS), we kindly ask you to submit a request to  views@pcr.uu.sePlease specify which monthly runs (versions/data releases) that you are interested in, and whether you seek only ensemble results or also constituent model results. The data can be provided in  .csv format and/or visualised. 
Replication data

REQUEST REPLICATION DATA

Due to the size of the data (>200 GB), replication data are currently only available for download via the CLI viewser (certificate required). Data can however be shared upon request to views@pcr.uu.se.
Please specify in your email which prediction model(s) your are seeking replication data for, as well as whether you seek the full replication dataset (>200 GB), or only segments thereof (e.g. specific time periods). Full datasets are sent as signed database dumps. They can be loaded into a PostgreSQL 9.6+ server, and will generate a new database that replicates the full work environment of ViEWS.  Segments of replication data are shared as .csv files. 
GIS RESOURCES

GEOGRAPHIC INFORMATION SYSTEM (GIS) RESOURCES

The VIEWS data are offered at two spatial levels of analysis: the PRIO-GRID and country levels. Both are presented at a monthly temporal resolution.  The material required to interpret and visualise data across these levels and resolutions can be found below. 
The country level
The list of countries included in the VIEWS datasets is derived from the  Gleditsch & Ward (1999) list of independent states. The geographic extents thereof (as well as the country IDs by which our data are presented) are determined by the GIS dataset CShapes v.0.6 ( Weidmann, Kuse & Gleditsch, 2010).  
List of the VIEWS-specific country IDs
Complete with corresponding ISO country names, numbers, and three-letter codes; Gleditsch & Ward country codes and other GW attributes; capital names with longitudes and latitudes thereof; dummy variables showing whether the given country is located in Africa or the Middle East; centroid longitudes and latitudes; and the calendar months for which the given country was/is used in the VIEWS datasets ( month_start and month_end). 
Shapefiles for the countries currently covered by the VIEWS data. 
The PRIO-GRID level
The PRIO-GRID cells used in VIEWS are derived from from PRIO-GRID 2.0  ( Tollefsen, Strand & Buhaug, 2012).  The latter is a standardized spatial grid structure consisting of quadratic grid cells that jointly cover all areas of the world at a resolution of 0.5 x 0.5 decimal degrees,  approximately 55×55 km around the equator.
List of PRIO-GRID IDs
Complete with the PRIO-GRID columns and rows in which they are located (from PRIO-GRID); latitudes and longitudes; as well as the corresponding VIEWS country IDs, Gleditsch & Ward country codes, three-letter ISO country codes, and country names.  
Month IDs
The VIEWS data are presented at a monthly resolution, identified by means of numeric month IDs. The month IDs take the form of a counter that start at 1 for January 1980, and increment by 1 for each calendar month thereafter. 
List of month IDs
List of the month IDs used in VIEWS. Complete with corresponding calendar months and years. 

Source code & documentation

Modeling code
Repository: viewsforecasting
Description:
Documentation of the prediction model currently in use (2022- ).
Prediction outcome:
Continuous (number of fatalities per country-month and PRIO-GRID-month).
License: CC-BY-NC

CitationHåvard Hegre, Forogh Akbari, Mihai Croicu, James Dale, Tim Gåsste, Remco Jansen, Peder Landsverk, Maxine Leis, Angelica Lindqvist-McGowan, Hannes Mueller, Malika Rakhmankulova, David Randahl, Christopher Rauh, Espen Geelmuyden Rød, Paola Vesco, 2022. ” Forecasting fatalities“. Report, Uppsala University, 2022-06-09.
Repository:  FCDO_predicting_fatalities
Description: 
A snap shot of the source code for the continuous prediction model as it looked when it was first released early 2022 ( the fatalities001 model) in collaboration with the UK FCDO. Further developments are pushed to the viewsforecasting repository.
Prediction outcome:
Continuous (number of fatalities per country-month and PRIO-GRID-month).
License: CC-BY-NC

CitationHåvard Hegre, Forogh Akbari, Mihai Croicu, James Dale, Tim Gåsste, Remco Jansen, Peder Landsverk, Maxine Leis, Angelica Lindqvist-McGowan, Hannes Mueller, Malika Rakhmankulova, David Randahl, Christopher Rauh, Espen Geelmuyden Rød, Paola Vesco, 2022. ” Forecasting fatalities“. Report, Uppsala University, 2022-06-09.
Repository:  views_competition
Description:
Code and data used in the clean-up and evaluation of the 2020 ViEWS prediction competition
License: CC-BY-NC

Citation: Vesco, Paola, Håvard Hegre, Michael Colaresi, Remco Bastiaan Jansen, Adeline Lo, Gregor Reisch & Nils B Weidmann (2022). “United They Stand: Findings from an Escalation Prediction Competition.” International Interactions 48(4). doi: 10.1080/03050629.2022.2029856
Repository:  OpenViEWS2
Description:
  Documentation of the prediction models in use between 2018 – 2021. Please note that these models are deprecated and thus no longer maintained. 
Prediction outcome:
Dichotomous (at least 25 BRDs per country-month, and at least 1 BRD per PRIO-GRID-month).
License: CC-BY-NC
Citation: Hegre, H., Bell, C., Colaresi, M., Croicu, M., Hoyles, F., Jansen, R., Leis, M. R., Lindqvist-McGowan, A., Randahl, D., Rød, E. G., & Vesco, P. (2021). ” ViEWS 2020: Revising and evaluating the ViEWS political Violence Early-Warning System“. Journal of Peace Research. doi:10.1177/0022343320962157
Repository:  OpenViEWS
Description:
Documentation of the prediction models in use between 2017 and early 2018. Please note that these models are deprecated and thus no longer maintained. 
Prediction outcome:
Dichotomous (at least 25 BRDs per country-month, and at least 1 BRD per PRIO-GRID-month).
License: CC-BY-NC
Citation: Hegre, Håvard, 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 RDichotomous (at least 25 BRDs per country-month, and at least 1 BRD per PRIO-GRID-month). weight: 400;”>ød, Gerald Schneider, Nina von Uexkull, and Jonas Vestby. “ViEWS: a Political Violence Early-Warning System”. Journal of Peace Research, 56, no. 2 (March 2019): 155–74. doi:10.1177/0022343319823860.
Data visualization
Repository: Mapper2
Description:
A VIEWS package that provides a set of tools to generate visualizations of the VIEWS data. Coupled with a set of Jupyter notebooks that illustrate different data mapping options. 
License: CC-BY-NC
API
Repository:  views_api
Description:
Documentation and user guide for the VIEWS API that allows retrieval of predictions and selected input data.
License: MIT
Data infrastructure
The repositories below contain all documentation and source code required to run the ViEWS prediction models that have been developed since 2017. Each prediction model is based on one of three different data infrastructures: ViEWS3 (current, see Hegre et al. 2022), ViEWS2 (2018-2021, see Hegre et al. 2021), or ViEWS1 (2017-2018, see Hegre et al. 2019). 
ViEWS3 (2022- )
Key repositories for ViEWS3
Repository: viewser
Description: 
A client package for interacting with the ViEWS3 cloud ecosystem. This library provides functions, as well as a CLI entrypoint that handles authentication and API calls, making it easy to seamlessly retrieve data, explore models, and much more. See the viewser  user documentation to learn more. 
License:
CC-BY-NC
Repository:  views3
Description: 
A repository that holds the main docker-compose file used to deploy ViEWS3. See the  wiki for how to run.
License:  CC-BY-NC
Remaining repositories for ViEWS3
Repository:  base_data_retriever
Description:
Database layer for ViEWS3
License: MIT
Repository:  queries_benchmark
Description:
A script to serve as a benchmark to ensure that ViEWS3 is working as expected. 
License: Default GitHub protection
Repository:  stepshift
Description:
A package that implements the stepshifting algorithm described in Appendix A of Hegre et al. (2020).
License: CC-BY-NC
Repository: viewserspace
Description:
A batteries-included workspace for writing modelling notebooks using viewser.
License: Default GitHub protection
Repository:  views_dataviz
Description:
A legacy library holding former data visualization functions used by the ViEWS team.
License: CC-BY-NC
Repository:  views_data_transformer
Description:
The transformation endpoint for resolving data queries
License: CC-BY-NC
Repository:  views_docs
Description:
A small service that proxies to various introspection routes to forward documentation to the user. The service also checks for optional verbose documentation, which can also be posted to expand upon what is already provided by the services.
License: MIT
Repository:  views_example_workspace
Description:
A repository that demonstrates how to work with the ViEWS3 cloud locally.
License: Default GitHub protection
Repository:  views_job_manager
Description:
The service responsible for launching and managing jobs.
License: MIT
Repository:  views_partitioning
Description:
A python package for partitioning data for the ViEWS project.
License: CC-BY-NC
Repository:  views_queryset_manager
Description: 
A  service that is responsible for managing and retrieving data pertaining to querysets, which are collections of links to remote data
License: MIT
Repository:  views_query_planning
Description:
A package that exposes a class that makes it possible to generate queries against a relational database using a network representation of the ViEWS database.
License:  MIT
Repository:  views_router
Description:
The router is responsible for routing requests to the right data source.
License: MIT
Repository:  views_runs
Description:
A package meant to help ViEWS researchers with training models by providing a common interface for data partitioning and stepshift model training. It also functions as a central hub package for other classes and functions used by ViEWS researchers.
License: CC-BY-NC
Repository:  views_schema
Description:
Contains data definitions used throughout the ViEWS 3 system.
License: MIT
Repository:  views_stepshift
Description:
Package that contains the modelling procedure from ViEWS2, along with helpers and auxiliary functions.
License: Default GitHub protection
Repository:  views_storefront
Description:
An nginx database that proxies requests to the rest of the system.
License: CC-BY-NC
Repository:  views_transformation_library
Description:
The data transformation library used by the views data transformer service.
License: CC-BY-NC
Repository:  views_storage
Description:
A package that contains various classes used for storage and retrieval of data within views.
License: MIT
Repository:  views_timecaster
Description:
Simple service for casting between time formats when querying data for ViEWS.
License: Default GitHub protection
Repository:  vgrouper
Description:
A service responsible for grouping collections of variables into querysets.
License: Default GitHub protection
Repository:  views_blob_storage
Description:
Blob storage manager, experimental solution for storing submitted data in a flexible but efficient way.
License: Default GitHub protection
ViEWS2 (2018-2021)
Repository:  OpenViEWS2
Description:
  Source code to run the 2018-2021 prediction models under the ViEWS2 data infrastructure, coupled with a  user guide to get started with ViEWS2. Please note that ViEWS2 is deprecated and thus no longer maintained. 
License: CC-BY-NC