How does SafeGraph define POIs? Just wondering if there are still many non-POI trips except the captured visit counts

Hi, I am using raw_visit_counts and normalized visit counts in the U.S. from the Monthly Pattern dataset. I am wondering how representative the visit counts can be for Americans’ total number of trips. Does SafeGraph also capture non-POI visits? What is the difference between POI and non-POI trips?

Hi @KANGLIN_CHEN_University_of_Florida , this is a good question. Have you seen SafeGraph’s Neighborhood Patterns? I think it addresses what you are asking and it is in the Dewey Marketplace.

Basically, Neighborhood Patterns tracks visits at the CBG level instead of the POI level. Depending on your use case, this could be helpful.

As for your last question on the difference between POI and non-POI trips, the POI visits are attributed based on clusters of device pings and POI geometries. In the Monthly Patterns (which are POI-based), a visit will only be attributed if the cluster of pings resides within the boundary of a POI (potentially with some buffer). So if there is a cluster of pings in the middle of nowhere with no POIs nearby, no visit would be attributed in Monthly Patterns.

1 Like