Dr. Nathan Wycoff (Massive Data Institute, Georgetown Univ.)

10/16/2023

Abstract: 

The 2022 Russian full-scale invasion of Ukraine forced millions of people to leave their homes and sent government officials scrambling to determine reasonable estimates of the number of people who would seek refuge in their countries in the early stages of the crisis. In this talk, we'll investigate the possibility of using various publicly available data to predict this forced movement. In particular, we establish Ukrainian-language insecurity and contextual indicators from data sources that include Twitter/X and Google Trends. We compare the usefulness of these indicators in predicting forced migration into three neighboring countries. To minimize the challenge of temporal misalignment between these organic data and actual movement, we develop a lagging and aggregation framework. We find that several of our indicators are qualitatively useful and that our Google Trends variables are a strong leading indicator of the observed forced migration. This highlights the potential for using a combination of publicly available organic data during emerging crises.