.svg)
Points of interest Data for GIS Research
A point of interest record fixes a single establishment to a coordinate and a category: a pharmacy, a park entrance, a charging station, a place of worship, sometimes with the footprint of the space it occupies. That turns the built environment into a layer that can be counted directly, so a researcher can measure what sits inside a quarter mile buffer or a block group rather than infer it from land use classes. Census and administrative files describe who lives in a tract; point of interest data describes what stands there, how it is classified, and when the listing opened or closed. Coverage varies sharply by category and by provider, which is why these databases are as often the object of study as the input to one.
How GIS researchers use points of interest data
Comparing and validating point of interest databases
Any spatial analysis that counts establishments inside a buffer inherits the coverage of whichever point of interest file supplied them, so GIS researchers routinely treat the database itself as the object of measurement, holding the study area fixed and varying the source to see how far a category count, a geocode or an open and close date moves. The comparison is usually run category by category, since agreement is highest for chain retail and lowest for small independent businesses, places of worship and informal community spaces. No source is a census of what exists, and a listing that persists after a closure is as consequential for a buffer count as one that was never recorded.
Accessibility and catchment modelling
Two step floating catchment and related accessibility measures need a supply layer of facility locations with categories, and point of interest data provides that layer at a finer grain than administrative facility registries, which list only licensed, funded or reporting sites. Researchers join those coordinates to block group population counts and a travel time surface to produce provider to population ratios that vary within a city rather than across counties. The weak point is that a point of interest record marks where a facility stands, not how many people it can serve or when it is open, so capacity has to be imputed from floor area, brand or staffing proxies.
Spatial equity in infrastructure siting
Mapping one class of facility against neighbourhood composition is the standard design for questions about who gets access to public infrastructure: geocoded station, clinic or store locations are joined to block group demographics and compared across income or racial composition deciles. Point of interest data makes the facility side of that join tractable, since it carries category, brand and coordinates for sites that no single administrative register covers completely. What it does not carry is whether a facility is usable, so a listed location that is out of service, gated behind a fee or restricted to residents will overstate access in precisely the places where access is contested.
Points of interest datasets available on Dewey
SafeGraph
Point of interest records with coordinates, category codes, brand and building footprint geometry for US establishments, the base layer for buffer counts, catchment models and joins to census geography.
View datasetData Axle
US consumer and business records that geocode to establishment level, useful as an independent second source for testing how far a buffer count or accessibility measure depends on which database supplied the locations.
View datasetEV Charging & Alternative Fuel Stations
Locations, connector types and network details for public electric vehicle charging and alternative fuel stations across the United States, a free point layer that joins directly to census geography for siting and accessibility work.
View datasetMore published work using points of interest data
Michael Hua · UCLA Electronic Theses and Dissertations · 2026
Roya Shomali, Nickolas K. Freeman, Gregory J. Bott, Iman Dayarian, Jason M. Parton · working paper · 2026
Youngjae Won · Cities · 2026
.svg)