AidData’s Financing the 2030 Agenda for Sustainable Development Dataset Version 1.0 provides aggregated data on Official Development Assistance (ODA) and Other Official Flow (OOF) commitments to the 17 Sustainable Development Goals (SDGs). This release builds on the previous iteration of the SDG coding methodology released in November 2017 (Estimating Baseline Aid to the Sustainable Development Goals). This release uses novel machine learning tools to automate this methodology and expand coverage through 2021.
AidData’s Autocoding Finance for Sustainable Development methodology is based on an analysis of ODA project descriptions and involves four steps: (1) a mapping between first-level activities and SDG targets; (2) training a text-classification tool leveraging three text-classification models on a dataset where this mapping has been hand-coded, then processing new data with this tool; (3) splitting an aid project across designated activities; and (4) splitting the dollar value of an aid project across the associated goals (for further detail, see Autocoding Finance to the Sustainable Development Goals Methodology, forthcoming, AidData). These steps allow us to estimate the total aid at the goal level for each of the SDGs.
The dataset was developed by AidData with the generous support of the William and Flora Hewlett Foundation.
A forthcoming version 1.1 data release will also include the full, unaggregated project-level data. However, as this dataset contains over 3.2 million project entries, we believe that a release of partially aggregated data will be most useful to the general research audience.