Risk Model: Forecasting Conflict Fatalities in Ethiopia Using ACLED Data

This project used time series analysis to forecast regional fatalities during the Ethiopian civil war using data from the Armed Conflict Location & Event Data Project (ACLED). I applied ARIMA models in R to predict monthly battle-related death tolls across the Tigray, Amhara, and Oromia regions.

The analysis revealed that fatalities were likely to increase in Amhara and Oromia, suggesting the conflict was shifting southward from Tigray — a critical insight for humanitarian response planning. Each region’s data was decomposed to examine trends, seasonality, and randomness, and model outputs were interpreted to inform resource allocation strategies for aid organizations.

Tools Used: R, ACLED data, Excel
Focus Areas: Conflict forecasting, humanitarian aid strategy, ARIMA modeling, geopolitical risk
Key Insight: Civil war violence was increasingly concentrated in regions outside of Tigray, offering a window for humanitarian access and forecasting operational hotspots.