Can we predict war using data and intelligent models? And how certain would we be in our predictions? A recent master research project by Oleksandr Zakotianskyi tries to answer these important questions…
“A new wave of violent conflicts around the world raises concerns about the security of the whole world. According to the Armed Conflict Location and Event Data (ACLED) Conflict Index, 12% more conflicts occurred in 2023 compared to 2022, and the amount of conflict increased by 40% compared to 2020, and this trend only increases with time. An underlying assumption and motivation of conflict early warning systems is that enhanced prediction and forecasting can better inform decision-making, reduce risk, and trigger more robust prevention and response measures from international actors.”
Why is this important?
“Policymakers are interested in uncovering the conflict dynamics and making robust forecasts to take anticipatory actions and reduce the impact on vulnerable people. At the same time, regular citizens are looking for accurate and not biased forecasts for their own security and awareness.”
The Model: Explainable and Transparent
The master project by Oleksandr Zakotianskyi provides the first publicly available and explainable early conflict forecasting model, capable of forecasting the distribution of conflict-related fatalities on a country-month level. The model is transparent, uses publicly available data and produces predictions up to 14 months into the future. The model implementation is based on the Natural Gradient Boosting framework.
The model and data preprocessing scripts are publicly available on GitHub.
The model presented by Oleksandr Zakotianskyi is based on the Natural Gradient Boosting framework developed by the Stanford ML group. This framework not only allows for predicting conflict-related fatalities but also provides an estimation of the conditional probability distribution for each prediction. Read more…
This master research project is part of the EURIDICE endeavor to develop new, societally relevant master education in the field of Digital Society and Global Citizenship, trying to tackle Grand Challenges in Society.
Download the full report.