Font Size: 

Today, centralised collection of national migration statistics allows a comprehensive analysis of migration patterns. The IPUMS database, for example, delivers a broad selection of migration surveys of 45 countries from 1960 to 2011. In total a massive sample of more than 350 million individuals and their migration behaviour can be assembled. However, such a large sample poses an ultimate obstacle to 'traditional' parametric statistical methods. Machine learning techniques, such as cart decision tree models, on the other hand, make it possible to reduce complexity and help to identify individual and contextual  characteristics related to individual migration.Initial findings monitor 'brain drain' phenomena on an international scale and suggest that migration could be driven climate change.