At the beginning of the month, Andrew Harper, UNHCR’s Director of Program Support & Management gave a speech at the re:publica conference in Berlin about the use of predictive analytics in humanitarian crises.
He presented three case studies:
- Using data to predict the likelihood and number of Syrian refugees moving across the Mediterranean Sea.
- Predicting future trends of asylum applications of Venezuelans by analysing social media, particularly Twitter, and using machine learning to assess intentions to flee.
- Somalia: Predicting and analysing population movements from Somalia by looking at a combination of weather data, market data and satellite images.
He doesn’t go into a lot of details regarding the use of social media data – it basically looks like a rather straightforward frequency analysis for certain terms, such as “flee” – but I still found it interesting. The talk is 27 minutes long, with the Venezuela case study starting at 11:40.
What are your thoughts? Please leave a comment below!