UNICEF Magic Box Dev Call to Action
Last month I wrote an article about an open source tool we started using at UNICEF Innovation, and the project we work on involving big data, AI, and data science called Magic Box. But, while it began with a human readable narrative, the mundane technical details I added probably kept most people from reaching the call to action at the bottom.
This article is for devs of any level. We’ll list some of the features we think will be useful at UNICEF, as well as for other humanitarian organizations. You can start by cloning this repository.
Most of the features below involve preparing datasets and connecting them to the client. If you have any questions, please chat with us on our public gitter channel.
Population
The first external dataset we connected to keplergl was shapefiles. To try it out:
- go to this demo, and click on “Select shapefile”
- select a country, and then a level, and click submit
In addition to simply showing the borders, we’d also like the option to color them according to population.
Two inputs are needed:
- Shapefiles — There’s no single source for the most accurate or agreed upon borders. gadm and hdx are to good collections. At the moment, we use gadm2.8, but would love to have 3.6 in addition to hdx. Here is a repo we last updated in April, 2017 to download them from gadm.
- Population rasters — Similar to shapefiles, there are many places to get pop rasters. We’ve used worldpop. To aggregate raster pixels by polygon, we’ve used postGIS and rasterstats.
What we’d like to see is any of the following:
- An open source repo for downloading and aggregating population rasters by shapefiles for every country.
- Population added as an option in the kepler-demo. On country select, shapefile and population should arrive separately, and be combined before being loaded to the map.
Health facilities
healthsites.io is a good source of health facility locations. Enable the fetching of health facilities by country as points from this API, and load them to the map.
