Drought insurance baseline survey in Ethiopia. Research coordination

Client: Oxford University, Centre for study of African Economics.

Improving loan targeting using poverty estimates from alternative data sources. is an online crowdfunding platform to extend financial services to financially excluded people around the world. Kiva invited the Kaggle community to help them build more localized models to estimate the poverty levels of residents in the regions where Kiva has active loans. The solution merged Kiva's household’s coordinates with clusters from the Demographic and Health Surveys (DHS). Granular and accurate poverty measures were obtained by using a k-nearest neighbor approach to match KIVA borrowers with DHS clusters.

Impact evaluation of micro insurance

Survey design, digital data collection with fieldwork supervision for the household survey fourth wave. Data cleaning and report

Client: Colombia University and Oxfam America.

Migration and Development Survey in Ethiopia: Survey design, data management and analysis.

Client: University of Maastricht

Recommender system identifying professionals that are most likely to answer career-related questions. is a nonprofit that crowdsources career advice for underserved youth. The final system was an hybrid of three recommenders (Student-to-Professional affinities, Content-based domain knowledge and Content-based questions similarities). The ranking obtained is then optimized with a Bayes model trained on professionals’ past interactions with the website.