An estimated 5.4 million Venezuelans live abroad, with over 220,000 residing in Brazil as of 2019. Many Venezuelans cross regularly between these countries for economic or family reasons, but border crossings in both directions were restricted as a result of the COVID-19 pandemic. As the Brazilian government considered reopening the border, UN Global Pulse teamed up with UNHCR in order to forecast the number of displaced persons who will enter Brazil from Venezuela and their need for humanitarian aid. This information is needed to determine whether the existing shelter infrastructure is sufficient or whether capacity must be expanded in anticipation of these new flows.
The team is experimenting with a variety of different methodologies, which include:
- Collecting real-time data on potential indicators from different data sources (e.g., Twitter, Google, Facebook, and radio) which might identify large numbers of people travelling to (or planning to travel to) the border.
- Applying machine learning and econometric forecasting methods to estimate future arrivals levels according to different economic and social indicators.
- Developing an interactive queueing simulation to model how new arrivals move through the different stages of border crossing.
The ultimate goal of the project is to evaluate the likelihood of different scenarios under consideration by UNHCR, each of which is associated with a different preparatory operational action. One of the core challenges to forecasting is that there is a large period of unusual data from the COVID pandemic, and since the border has remained fully or partially closed there is not a clear expectation about what the current demand for crossing is. Therefore, the team is testing multiple different strategies for predicting future outcomes, which it will continue to update as more data becomes available.