By Andrea Pellandra and Alejandra Moreno Ramirez, UNHCR’s Global Data Service
In April 2025, UNHCR and the OECD-working in partnership with the International Data Alliance for Children on the Move (IDAC)-launched the first-ever OECD-UNHCR Datathon. The initiative brought together students, researchers, and data practitioners to explore one central question: How can data be used to better understand and respond to the needs of forcibly displaced and stateless children?
Participants leveraged datasets from the UNHCR’s Microdata Library (MDL), which contains data on forcibly displaced and stateless populations from more than 70 countries, to generate actionable, policy-relevant insights that could inform humanitarian and development responses.
Turning Data into Impact
Participants worked in multidisciplinary teams, drawing on a curated selection of MDL datasets and integrating any other relevant data openly available in other repositories where needed. Guided by three thematic challenges, teams explored how data can shed light on the experiences of children affected by displacement. One challenge focused on identifying data gaps and improving the inclusion of forcibly displaced and stateless children in existing datasets. Another challenge examined disparities and commonalities between displaced and host community children across areas such as education, health, and living conditions. The third challenge encouraged participants to analyze how multiple vulnerabilities-such as legal status, poverty, and access to services-intersect, and to propose holistic solutions aligned with the Sustainable Development Goals.
The Datathon encouraged participants to pair rigorous analysis with compelling storytelling. Final submissions were evaluated based on a combination of metrics ranging from data science to visual narratives, and policy insights that gauged their ability to translate complex findings into practical, real-world relevance.
Recognizing Excellence: The Winning Teams
After evaluation by a panel of experts from UNHCR, OECD, and UNICEF, two teams were selected for their innovative approaches, analytical depth, and potential to influence real-world decisions.
The first winning team, Team Craic, composed of William Paja, Matt Murtagh-White, and Wooyong Jung, developed The UNified Model – Predicting Education Outcomes for Displaced Children in Data-Scarce Contexts. Their project addressed the challenge of missing or incomplete education data by building a machine learning model that combines household-level microdata with geospatial information such as proximity to schools, healthcare facilities, and conflict zones. Focusing on Iraq and Uganda, the team used LightGBM algorithms to identify key predictors of school enrollment—such as age, household income, distance to services, and child labor—and computed SHAP values for interpretability. The team also designed a scalable version of their model that could generate district-level predictions in areas without available survey data. Tested on conflict-affected regions in Iraq using data from the Iraq Multi Cluster Need Assessment (2021) and validated against the Uganda Joint Multi-Sector Needs Assessment (2018), the model demonstrates robust performance and offers a practical, adaptable tool for enhancing education targeting in humanitarian contexts where traditional data collection methods are constrained.
The second winning team, Data for Hope, composed of Eyram Espoir Tetshie and William Kokou Amedanou, focused on layered vulnerabilities affecting child health in refugee contexts. Their project, Multidimensional Vulnerabilities and Child Health in Uganda, used data from the Uganda Joint Multi-Sector Needs Assessment 2018 to create a composite health index based on indicators such as vaccination coverage, mosquito net usage, and recent illness. The team constructed additional indices for household-level access to WASH, food assistance, shelter quality, school attendance, and legal documentation. Using both regression analysis and machine learning models, they found that children’s health outcomes were shaped by the interaction of multiple factors—poor shelter and inadequate hygiene, for instance, were especially harmful when compounded by food insecurity. Their findings point to the need for integrated, multisectoral approaches that can address the complex and overlapping risks faced by children in displacement.
On the Global Stage
Both winning teams have been invited to present their work at the upcoming 4th International Forum on Migration Statistics (IFMS) in Malmö, Sweden, from 16 to 18 June 2025, and at the next IDAC conference. These platforms will give the teams an opportunity to share their findings with a global audience of policymakers, data experts, and humanitarian actors.
Register to attend IFMS 2025 online or in person: https://ifms2025.org/register.
Looking Ahead
The OECD-UNHCR Datathon showcased what is possible when data is used not just to describe problems, but to find solutions. From scalable education models to multidimensional health analysis, the projects highlighted the value of open data and cross-sectoral collaboration in supporting displaced children worldwide.
UNHCR remains committed to strengthening data ecosystems that enable humanitarian response, inform protection efforts, and promote inclusive, evidence-based policies. We extend our sincere thanks to all who contributed to the success of the Datathon-and look forward to continuing this important work together.