Applications: This can provide baseline information to understand the exposure of employment to various hazards, and/or to conduct accessibility analyses to employment. These layers can also be used as inputs into quantitative urban models or for spatial form analyses (employment clustering, average distance to jobs). These predicted job layers have been succesfully produced for eight different task teams, covering a total of 27 cities in 12 countries since June 2022.
Data requirement: Global data: layers available in Open Street Maps (OSM) and Google Earth Engine
Final deliverables: A GIS layer (shp, geopackage) that provides the predicted spatial distribution of jobs in the selected urban areas using proxy data.
Geographic scope: City-level
Indicative Cost / working time: ~$1.5k for the 1st city - 2 days; with scale economies and therefore lower costs from 2nd city onwards
Methodology peer-reviewed, cleared (date)?: Yes - methodology was reviewed and published in the PRWP series in March 2022. A short demonstration note and a blog post decribe the algorithm, its purpose and its performance in less technical terms
Time Required for delivery: 1 - 3 weeks (depending on other commitments)
Application of a Machine Learning algorithm to predict for the spatial distribution of jobs within urban areas. The algorithm has been successfully trained on and validated against employment data in 14 cities of the developing world.
