Predicting asylum-related migration flows with machine learning and large-scale data – Global

EUAA and European Commission scientists unveil big data-based asylum migration forecasting model

On January 27, 2022, researchers working for the European Union Agency for Asylum (EUAA), the European Commission (Joint Research Centre) and the University of Catania published a new methodology for forecasting asylum applications. asylum lodged in the EU, based on machine learning and big data.

Published in Nature Scientific Reports, the world’s 6th most cited journal, the model’s aim is to increase the preparedness of EU Member States for a sudden increase in asylum applications in order to process them quickly and fairly; while providing adequate reception conditions in accordance with EU law.

By integrating traditional administrative data on migration and asylum, such as detections of illegal border crossings and recognition rates in destination countries, with big data on negative and conflict events, as well as internet searches in the countries of origin; this new machine-learning system, known as DynENet, can predict EU asylum applications up to four weeks in advance.

The approach draws on migration theory and modeling, international protection and data science to deliver the first comprehensive asylum claim forecasting system based on adaptive models and large-scale data. Importantly, the approach can be extended to predict other social processes.

Since 2011, the EU, initially through the European Asylum Support Office (EASO) and since January 2022 with its new European Union Agency for Asylum (EUAA), has been assisting its Member States members to establish the only multinational asylum system in the world. Combined with an enhanced mandate to provide operational support to Member States under pressure, the data provided by this new forecasting tool could not only help Member States increase their internal preparedness, but also indicate where frontline operational assistance from the Agency to national authorities in the EU may be necessary.

The research paper can be viewed at: Forecasting asylum-related migration flows with machine learning and data at scale (nature.com)

Sherry J. Basler