AJ Friend
AJ is a data scientist at Waymo, formerly a data scientist at Uber Eats and maintainer of H3 and H3-py.
Sessions
We define an aspect ratio for spherical point sets to compare DGGSs,
complementing metrics like area variance or isoperimetric quotient. We share
a fast open-source implementation, survey aspect ratio distributions across
grids, and discuss generalizing to the ellipsoid.
We can select sets of DGGS cells meeting some criterion (like travel
time for isochrones) while staying compactible to reduce storage.
We pose an optimization problem trading off the two objectives,
give examples across DGGSs, and extend to co-compaction.
Compaction reduces the storage cost of cell sets in hierarchical DGGSs. We
cover compaction along with efficient set membership querying, why the "gaps" in
mixed-resolution sets are merely an illusion, and a new co-compaction algorithm for
additional savings.