Forecasting the onset of conflict at a fine-grained temporal level has important implications for early-warning systems and can provide timely information to policymakers, humanitarian organisations, and peacekeeping actors. Typical conflict prediction methods rely on variables such as political instability, economic conditions, population characteristics, or historical conflict patterns. However, this data is often relatively coarse in the temporal domain, making it difficult to detect rapidly developing changes and provide timely predictions of conflict outbreak. Automatic Identification System (AIS) data, which records the movements and locations of vessels over time, offers a potentially complementary source of information that is both highly granular and rapidly updated. Changes in maritime activity, shipping routes, port visits, vessel density, or patterns of movement may provide an early signal of disruptions or deteriorating security conditions before these are captured by more conventional predictors. Building on existing work in conflict forecasting, this project should investigate whether AIS-derived features can improve the early detection and forecasting of conflict onset compared with more conventional conflict predictors. In particular, the project could examine whether the fine-grained and reactive nature of AIS data enables models to identify emerging conflicts at higher spatial and temporal resolution, potentially providing earlier warnings of conflict outbreaks. A working knowledge of the programming language Python or R, statistical modelling, and machine learning is strongly advised to work on this project. All interested students must first email c.fritz@tcd.ie to arrange a short, informal meeting to discuss the project.
This project is in collaboration with Prof. Thomas Chadefaux from the School of Political Science at TCD