During the development of Piccard (https://github.com/fabioasdias/piccard), one of the issues raised by a stakeholder was the option to upload new data, possibly with a new geometry.
Mixing geometries makes everything significantly more difficult. With Piccard, we removed the need for geographic hamornization, but the variables themselves were carefully curated to match across time, involving previously defined geometries.
This project aims to remove that need, allowing the user to upload data in whatever format/geometry/convention while combining it with previously uploaded information. This is far from done, computational geometry is challenging by itself, when mixed to data science, it doesn't get any easier.
This research was supported by a University of Toronto Connaught Global Challenge grant and is part of the Urban Genome Project.