by Moses Sitati, HDX Data Lab Manager and Luis Capelo, HDX Data Scientist
The Humanitarian Data Exchange (HDX) team established a Data Lab in Nairobi, Kenya in late 2014 to offer data services to partners and to connect data systems from across the region. One of our first projects was a data hunting exercise on behalf of the OCHA Regional Office for Eastern Africa. The data would be used to better understand people’s vulnerability, in line with the INFORM risk model.
The original brief was to collect data for 90 indicators at the county level (admin 2) across 10 countries. After some refinement, this was narrowed down to 42 indicators at the state level (admin 1) across 10 countries. The countries were Burundi, Djibouti, Eritrea, Ethiopia, Kenya, Rwanda, Somalia, South Sudan, Sudan and Uganda.
The data covered topics such as livelihoods, health conditions, infrastructure, education, and population displacement. The sources for the data included Government ministries, UN agencies, the World Bank, and the private sector.
Over a three month period, we conducted physical and virtual outreach in search of the data. Data was manually extracted from static maps, GIS files, PDF documents, and Excel spreadsheets. Overall, about 25 percent of the data was considered complete, meaning there was data at the national and sub-national levels for every state in a specific country. No single indicator was available across all 10 countries at the sub-national level. (See the heatmap with countries and indicators).
The exercise showed the challenges contributing to the data-deficit in the developing world: data is either not being collected or when it is, it is not shared in accessible formats. These findings are in line with the data availability for indicators included in the Millennium Development Goals, as reported by the UN expert’s group in A World That Counts.
The lack of data has implications for gaining an accurate understanding of regional vulnerability. These caveats need to be made clear when the analysis is prepared so that decision makers don’t mistake an impression with reality. At the same time, new data sources, such as sensors and mobile phones, can be explored to fill some of these gaps. A recent article in The Economist on Africa’s ‘missing data’ explores these issues further.
Using the data that was available, OCHA Eastern Africa worked with the Intergovernmental Authority on Development (IGAD) to release a regional analysis for the greater Horn of Africa. The initial results show that the highest humanitarian risk is present in the border areas due to high exposure to climate shocks, low levels of development, and the presence of people displaced by conflict.
The Data Lab team continues to search for more detailed data on the borderlands region – a geographic strip running from the Karamoja region in north-eastern Uganda to the Gedo region in southern Somalia. The initial borderlands dataset is available on HDX.