These Are The Photos in Your Neighborhood
Neighborhood boundaries are a fluid thing. My definition what encompasses my neighborhood (Multnomah Village in Portland) may vary greatly from the person across street.
It’s also one of the Holy Grails to online search. The ability to confine a search to a particular neighborhood. Roost does this particularly well, as does Estately and to a lesser degree Zillow.
Zillow also took the step of creating a set of neighborhood boundaries (over 7000 of them) that they released earlier this year in Shapefile format.
Another effort is slowly taking shape (literally) however. The popular photo-sharing site Flickr is also releasing Shapefiles of neighborhoods, using the data it collects from the geotagged photos in its database to accumulate a collective definition of what makes a neighborhood.
The project, which is in early Alpha, can best be viewed on the mashup site Neighborhood Boundaries. Simply punch in your neighborhood name (or city name) and see how the boundaries are being defined in your neck of the woods.
Again, this is early Alpha, so many markets won’t have much coverage. But the possibility of user-defined neighborhood boundaries is intriguing to say the least.
If you enjoyed this post, make sure you subscribe to my RSS feed!
Make sure to follow Inman News on Twitter too!
Matt Goyer - Redfin | Nov 19, 2008 | Reply
Interesting data source find. Especially considering how they’re generating the outlines!
At Redfin, in order to provide a high quality user experience we mashed together three data sources because we couldn’t find one that provided both adequate coverage and was high quality. This took an extensive amount of time shaping, buffering, simplifying our new data set to get them “right.”
To me a great neighborhood initiative would provide an initial set of neighborhoods and then would use community and developer feedback to iterate on them so that they could be consumed by many sites without time consuming manual processing.
tony - forsalebylocals | Nov 20, 2008 | Reply
interesting development on being able to change the neighborhood shapefiles. Although Zillow does have neighborhood shape files, if I recall correctly they opted for the most restrictive open license which means no changes or additions.
I agree with matt in the comment above. The best data would be a crowdsourced iterative approach.
Drew Meyers from Zillow | Nov 20, 2008 | Reply
Tony-
The Zillow neighborhood files are available under a creative commons share alike license. Anyone can modify the files at will so long as they make the updates available to everyone else to use. We think that’s a pretty fair deal, but if there is some specific project that you feel you can’t take on as a result of that license, please let me know.
If there are people who have ideas as to the best approach to crowdsource a complete and accurate nationwide set of neighborhood boundaries, I’d love to hear from you. I think a standard dataset would open up the gates to all sorts of mashups with everyone being able to tag their data to the same boundaries.
Matt Goyer - Redfin | Nov 20, 2008 | Reply
re: I think a standard dataset would open up the gates to all sorts of mashups with everyone being able to tag their data to the same boundaries.
Yes, when two sites have different definitions of neighborhoods it then makes it difficult for the sites to partner on a neighborhood feature. This is something we’ve run into and been frustrated by.
Pete Goldey - Onboard Informatics | Nov 20, 2008 | Reply
What you are calling for is exactly what we do. We collect boundaries from many sources (ranging from end users to government and other agencies in every market) and provide a consensus shape and definition.
Redfin is one of our clients actually. (Matt if you have any specific changes you’d like to see, please let us know)
We do NOT start with the crowd / cloud however. Doing so results in the issues folks have found with Zillow’s and other crowdsourced boundaries: there needs to be some process around how they are created and modified.
Most neighborhoods DO have pretty well defined areas – but it takes time and effort to sift the wheat from the chaff. We’re covering 45k – 50k of them at the moment in the US with several thousand more in Canada. We do not let end users modify the shapes directly, but rather solicit feedback which we then validate. Otherwise you get “wishful thinking” set of neighborhoods rather than any semblance of accuracy in your dataset.
The flikr version would suffer from mis-tagging issues by end users. We’ve tried similar approaches and found that without constant attention from a knowledgeable team, you end up with a lot of garbage.
Controlled influence of crowds is closer to how we work.
Depending on what your goal is (”show me pictures from a neighborhood” vs. “show me average home sale prices or tax values from a neighborhood”), your tolerance for fuzziness is likely to be quite different.
- Pete
Overland Park Real Estate | Nov 21, 2008 | Reply
Good information, thanks for the find Joel.