Do changes from release to release for places, buildings reflect real-world changes?

[robfitzgerald]

Hi there! We are evaluating a move to OvertureMaps as a points-of-interest dataset for our access model, the MEP Metric. We know that, between releases, Places datasets can be updated with additional data sources from time-to-time. But real-world changes in points of interest are likely also captured from release to release. We are looking for a workflow to generate our metric at some cadence that can capture real-world changes in the opportunity space without drowning those changes in false-positives.

As it stands, there doesn't seem to be a way in the Places or Buildings schema to separate row creation date (i.e., version) and real-world date data, such as the earliest-known date a place was observed open for business. Is the latter something that may come in later schemas or is it not really feasible?

Thank you for all of the great work on OvertureMaps! 🚀

[cghoehne]

To add to Rob's point: our goal is to understand, as new releases come out, is there any way to control for increased data coverage vs increased real-world development (i.e., changing land use over time)? Our metric measures accessibility to places, so we are interested in the ability to measure over time how access to specific kinds of opportunities is changing across different regions. But we are unsure at this stage how OvertureMaps is adding new real-word developments vs filling in missing coverage of existing places/buildings with new releases.

I am sure it is probably difficult to determine the original real-world creation or opening date of a place or building. But it would be helpful for us to understand the process that leads to new additions in Places and Buildings datasets so we can judge if we can attribute increases in access to certain places in one city or another to new additions across releases.

E.g., say we ran an analysis using OvertureMaps Places and saw Austin and Denver had a 2% increase in presence of restaurants. How can we disentangle if this is due to new developments (a net increase in restaurants opening) versus these previously existing but were not captured in a previous release?

[danabauer]

Thanks @robfitzgerald and @cghoehne. These are fantastic questions. As you've noticed, Overture has been adding new data sources to places and buildings to increase coverage and improve quality for those themes. We do track in the first_seen column in our GERS registry when Overture first became aware of a entity, but as you pointed out this does not necessarily mean the entity is a newly-constructed building or newly-launched business, and it doesn't capture a real-world opening date.

There are a few ways you might enrich Overture data with indicators of new development. In Denver, for example, you could match Overture places and buildings with the locations of permits for new construction or licenses for new types of businesses, pulled from Denver's Open Data Catalog. Austin has its own open data portal with similar datasets. (I'm just using the two cities you mentioned as examples.) We have a lengthy tutorial that shows you how you to match and "GERS-ify" datasets.

We're also working on improving our schema tooling and schema documentation to make it easier for folks to build their own schema extensions. These improvements are on our product road map for later this year.

Hope this helps!

[robfitzgerald]

@danabauer thanks for the prompt reply! and this is great advice, since as you suggested, we are for the most part looking to perform this analysis in a few metro regions. in the long run, we also hope to publish national-scale MEP geoparquet datasets at some set release cadence to capture year-over-year changes. we will try and keep an eye out for any community efforts to append this kind of date information to the Places and Buildings datasets as we continue our process of integrating OvertureMaps into our pipelines. thanks! 👋

[danabauer]

Glad to hear it!