@ross_epstein_safegraph has Safegraph, or anyone in the community, worked on an audit of safegraph POI coverage compared to a local source, like business permits? Say, whether there are certain types of businesses or locations that are more likely to be missing from Safegraph Core Places?
Some federal sources (QCEW, I believe) contain counts of establishments by county. I’ve often thought it would be a useful metric to create – not only for an audit but as a way to re-weight when doing various geographic aggregations.
Good idea @Aaron_Flaaen_Federal_Reserve_Board. My main worry about this approach is being able to distinguish establishments that should be “POIs” in the sense of people physically visiting, vs. just general businesses which may include entities with no brick-and-mortar presence.
@Derek_Ouyang_Stanford this is a good idea. We’re always interested in benchmarking our POI coverage by category against “truth sets,” but truth sets are pretty elusive, and to your point, our scope only includes physical locations. We’re ramping up our ability to evaluate large datasets (including gov. data), so we may have some rough stats to report soon. The goal is to identify the top categories where we have weak coverage and then search for sources that give us the most lift in these weaker categories.