Agent-based modelling of illegal fishing

The problem

Illegal, unreported and unregulated fishing is a classic complex-systems problem: many actors, partial observation, and incentives that shift the moment enforcement does. This commissioned engagement asked a simple question with a complicated answer: given what can be observed, where should limited enforcement effort go?

The contribution

I built an agent-based model of the fishery, in which individual vessel agents decide where to fish, whether to comply and how to respond to patrols, based on behavioural rules grounded in the available data — so patrol strategies could be explored in silico, including displacement effects: the tendency of illegal activity to move rather than stop.

A large share of the work was an honest data pipeline: mixed-resolution vessel tracks, gaps that are themselves informative, and ecological covariates on entirely different grids, with the uncertainty carried through to the model rather than swept away.