ViEWS2020: Revising and evaluating the ViEWS political Violence Early-Warning System
| Author | Remco Jansen,Espen Geelmuyden Rød,Mihai Croicu,Angelica Lindqvist-McGowan,Paola Vesco,Frederick Hoyles,David Randahl,Curtis Bell,Michael Colaresi,Maxine Ria Leis,Håvard Hegre |
| Published date | 01 May 2021 |
| Date | 01 May 2021 |
| DOI | 10.1177/0022343320962157 |
| Subject Matter | Special Data Features |
Special Data Features
ViEWS
2020
: Revising and evaluating
the ViEWS political Violence
Early-Warning System
Håvard Hegre
a,b
, Curtis Bell
a,c
, Michael Colaresi
a,d
, Mihai Croicu
a
, Frederick Hoyles
a
,
Remco Jansen
a
, Maxine Ria Leis
a
, Angelica Lindqvist-McGowan
a
, David Randahl
a
,
Espen Geelmuyden Rød
a
, and Paola Vesco
a
Abstract
This article presents an update to the ViEWS political Violence Early-Warning System. This update introduces (1) a
new infrastructure for training, evaluating, and weighting models that allows us to more optimally combine con-
stituent models into ensembles, and (2) a number of new forecasting models that contribute to improve overall
performance, in particular with respect to effectively classifying high- and low-risk cases. Our improved evaluation
procedures allow us to develop models that specialize in either the immediate or the more distant future. We also
present a formal, ‘retrospective’ evaluation of how well ViEWS has done since we started publishing our forecasts
from July 2018 up to December 2019. Our metrics show that ViEWS is performing well when compared to previous
out-of-sample forecasts for the 2015–17 period. Finally, we present our new forecasts for the January 2020–
December 2022 period. We continue to predict a near-constant situation of conflict in Nigeria, Somalia, and DRC,
but see some signs of decreased risk in Cameroon and Mozambique.
Keywords
Africa, armed conflict, ensemble modeling, forecasting, model criticism
Overview
This article presents an update to the ViEWS political
violence early-warning system first presented in Hegre
et al., 2019. We outline improvements to a number of
components: we have enhanced ViEWS’ abilityto forecast
conflict onsetsand to separate low- and high-risk cases; we
have made adjustments to the dependent variables to
increase the usefulness of the system, improved the meth-
odology, andexpanded the set of predictors.We first sum-
marize and motivate these changes, and proceed to show
how these revisions improve performance. The revisions
primarily pertain to forecasts at the country level. Changes
to the subnational level have been more incremental, and
we therefore allocate less space tothese developments.
In line with ViEWS’ goal of maximal transparency, we
also revisit ViEWS forecasts published in Hegre et al.
(2019) and the monthly updates on the website (https://
pcr.uu.se/research/views/current-forecasts/). The evalua-
tion shows that overall predictive performance is in line
with our expectations in Hegre et al. (2019). Finally, we
summarizethe new forecastsfor the January 2020–Decem-
ber 2022 period.
1
aDepartment of Peace and Conflict Research, Uppsala University
bPeace Research Institute Oslo (PRIO)
cUS Naval War College
dUniversity of Pittsburgh
Corresponding author:
havard.hegre@pcr.uu.se
1
A set of Onlineappendices, availableat http://views.pcr.uu.se,provide
additionalinformation, and showresults for all three forms of organized
violence recordedby the Uppsala Conflict Data Program (UCDP).
Journal of Peace Research
2021, Vol. 58(3) 599–611
ªThe Author(s) 2021
Article reuse guidelines:
sagepub.com/journals-permissions
DOI: 10.1177/0022343320962157
journals.sagepub.com/home/jpr
To continue reading
Request your trial