OUR RESEARCHThe GeoQuery project is more than just a tool for data access - we also promote and conduct interdisciplinary research to better enable the use of spatial data in decision making. Research topics focus on four themes:
- Bridging the gap between machine learning, econometrics and geography to better understand the conditions under which aid is effective (or, not).
- Computational optimization for the processing and creation of large-scale geographic datasets.
- Novel methods for the accurate modeling of aid impacts under conditions of uncertainty or missing data.
- Applied case studies examining the impact of geographically-located interventions.
MACHINE LEARNING & AIDAdopting tools recently developed in silicon valley to understand the conditions under which product changes lead to better monetary outcomes, we are conducting global analyses to examine when and under what geographic conditions aid is most and least effective.
- Zhao, J., Kemper, P., Runfola, D., 2017. Quantifying Heterogeneous Causal Treatment Effects in World Bank Development Finance Projects. Data Mining and Knowledge Discovery (ECML PKDD)
- Runfola, D., Ariel BenYishay, Jeffery Tanner, Graeme Buchanan, Jyoteshwar Nagol, Matthias Leu, Seth Goodman, Rachel Trichler and Robert Marty. 2017. A Top-Down Approach to Estimating Spatially Heterogeneous Impacts of Development Aid on Vegetative Carbon Sequestration. Sustainability 9(3), 409. doi:10.3390/su9030409.
- Zhao, J., Runfola, D., Kemper, P. 2017. Simulation Study in Quantifying Heterogeneous Causal Effects. WSC Proceedings. Available online at: http://informs-sim.org/wsc17papers/.
COMPUTATIONThe spatial datasets we provide and use in our analyses can frequently consist of trillions of pixels of data. To process this data accurately and quickly, we have expanded the capabilities of a number of existing computational geography approaches to work effectively in high performance cluster computing environments.
- Goodman, S., BenYishay, A., Runfola, D., 2017. Overview of the geo Framework. AidData. Available online at http://geo.aiddata.org/. DOI: 10.13140/RG.2.2.28363.59686
- Goodman, S. Updates and Modifications to the RasterStats Python Module. Pull 144, 146, 148. https://github.com/sgoodm/python-rasterstats
UNCERTAINTYMore often than not, data has error. Surprise! This is particularly true of spatial data, as (for example) the location of international aid delivery is not always known with perfect accuracy. We develop new methods to overcome these and related to challenges to the use of spatial information in impact evaluation.
- Marty, R., Goodman, S., LeFew, M., BenYishay, A., Runfola, D. 2016. Modeling AidData Using geo(query). AidData. Available online at http://geo.aiddata.org/docs/ .
- Runfola,D., BenYishay,A. ,Goodman,S., 2016. geoMatch: An R Package for Spatial Propensity Score Matching
- Runfola, Daniel, Robert Marty, Seth Goodman, Michael Lefew, and Ariel BenYishay. 2017. geoSIMEX: A Generalized Approach To Modeling Spatial Imprecision. AidData Working Paper #38. Williamsburg, VA: AidData. Accessed at http://aiddata.org/working-papers.
APPLICATIONSThe best way to figure out the limits of your data and methods is to go out and try to use them alongside decision makers in the real world. Most of our methods either are actively being refined or have been edited in the past due to prolonged engagements with a number of different international aid organizations.
- Batra, G., Anand, A., Goodman, S., BenYishay, A., Nyoteshwar, J., Runfola, D., 2017. A Value for Money Analysis of GEF Interventions in Land Degradation and Biodiversity, http://www.gefieo.org/evaluations/value-money-analysis-gef-interventions-land-degradation-and-biodiversity
- Runfola, D.M., BenYishay, A., Buchanan, G., Nagol, J., Tanner, J., Valuing Environmental Impact of World Bank Projects: A Case Study of Integrating Value for Money Analysis into Impact Evaluations. What Works? Value for Money in Impact Evaluations. Independent Evaluation Group, the World Bank.