Michal Migurski’s contributions to the open data community are almost too numerous to count. Most recently, Mapbox has been pleased to support his work on the OpenAddresses project, including more thorough evaluation of its progress, which he discusses here.
OpenAddresses, the free and open global address collection, turned three years old this year. Address data is critical infrastructure for basic mapping tasks, like finding a location to get driving directions or measure distances. Mapbox’s geocoder uses address data from a number of sources, including OpenAddresses. This community open data effort to collect authoritative address points from around the world recently reached over 440 million records with substantial coverage for dozens of countries. Mapbox has been supporting OpenAddresses since day one, and this spring I’ve been lucky to get to work on detailed address coverage data here in the San Francisco office.
OpenAddresses (OA) is a freely-licensed data set, so unlike many proprietary datasets that come with contractual restrictions, it’s possible to talk about its data quantity and quality in great detail. Now, we’ve put together a detailed summary of how OA’s global addresses compare to population density from NASA SEDAC’s Gridded Population of the World (GPW) dataset, as well as a regularly-updated page with country-level coverage data:
OA address density compared to worldwide population.
| Country | Coverage Group | Land Area Covered | Population Covered | Addresses per 1,000 People | |||
|---|---|---|---|---|---|---|---|
| 🇦🇺 | Australia | Complete | 100% | 7,741K km² | 100% | 23,948K people | 847 |
| 🇦🇹 | Austria | Complete | 100% | 83K km² | 100% | 8,551K people | 328 |
| 🇧🇲 | Bermuda | Complete | 100% | 126 km² | 100% | 57K people | 335 |
| 🇧🇷 | Brazil | Substantial | 100% | 8,500K km² | 99% | 206,104K people | 403 |
| 🇨🇿 | Czech Rep. | Complete | 100% | 79K km² | 100% | 10,560K people | 268 |
| 🇩🇰 | Denmark | Complete | 100% | 45K km² | 100% | 5,660K people | 641 |
| … | |||||||
| 🇨🇦 | Canada | Substantial | 16.4% | 1,576K km² | 31.5% | 11,307K people | 633 |
| 🇺🇸 | United States | Substantial | 58.2% | 5,435 km² | 78.9% | 254,082K people | 609 |
| … | |||||||
A sample of OA coverage data for May, 2017. See an updated table.
GPW is a fascinating dataset hosted at Columbia University, featuring estimates of world population consistent with United Nations counts for the past 20 years in a 1km×1km raster grid. We can use it to estimate the number of addresses per 1,000 people for countries where we have even partial coverage. For example, OpenAddresses completely covers as many as 21 countries, and almost 80% of the population of the United States.
Yellow polygons show administrative areas where OA data sources exist in Europe.
Dark green shows dense address availability in Australia and New Zealand, where OA has countrywide address sources from LINZ and the government of Australia.
OA provides a space to collaborate on Github, where a bustling community of participants seek out new sources of address location data from government and local authorities. If you’d like to get involved in OpenAddresses, there are plenty of ways to help:
- Check out open issues for ongoing work.
- If an authority in your area is missing from OA, reach out to ask for data.
- Know of other open data sources? Submit them here!