Correlations and forecast of death tolls in the Syrian conflict

TL;DR

This paper analyzes the temporal and spatial correlations in Syrian conflict death tolls, revealing patterns that improve forecasting of future casualties and understanding conflict dynamics.

Contribution

It introduces a method to analyze correlations in death toll data and demonstrates how this improves prediction accuracy during the Syrian conflict.

Findings

Death tolls follow a log-normal distribution.

Strong correlations exist between cities and consecutive days.

Death events are predictable based on observed data.

Abstract

The Syrian civil war has been ongoing since 2011 and has already caused thousands of deaths. The analysis of death tolls helps to understand the dynamics of the conflict and to better allocate resources to the affected areas. In this article, we use information on the daily number of deaths to study temporal and spatial correlations in the data, and exploit this information to forecast events of deaths. We find that the number of deaths per day follows a log-normal distribution during the conflict. We have also identified strong correlations between cities and on consecutive days, implying that major deaths in one location are typically followed by major deaths in both the same location and in other areas. We find that war-related deaths are not random events and observing death tolls in some cities helps to better predict these numbers across the system.