Data & Methodology

DeCODE is informed by data from over 3,700 projects across the World Bank spanning the last 20 years. Using advanced text mining techniques, DeCODE identifies delivery challenges faced in each of these projects and predicts potential challenges for future projects.

Data Vintage

Data in DeCODE includes projects that concluded more than circa 20 years ago, (i.e., prior to 1995). The bulk of the data comes from projects that concluded after 2000. Currently, DeCODE uses reports from closed projects to identify challenges. In an upcoming iteration, DeCODE will also utilize status reports from active projects to predict future challenges.

Sectors

To facilitate a cross-organizational system, DeCODE uses sectors that can be mapped to every organization's business taxonomy. Fifteen sectors have been identified in all. Projects with multiple sectors have been weighted appropriately.

Countries and Regions

Overall, DeCODE has more than 250 projects for each of the six global regions. The majority of the projects are concentrated in the Latin America and Caribbean region, and in Sub-Saharan Africa.

Predictions

DeCODE uses the Bayesian probability of occurrence to predict future delivery challenges in a given sector or country. The team is currently building other robust measures to incorporate the level of impact of every given challenge.

For questions, comments, or suggestions contact us.