In the fall of 2013, we set out to expand Niche’s coverage beyond education to cover every city, town, and neighborhood in the United States.
Niche Local Beta
In April 2014, we launched the public beta of Niche Local, and it is very much a “beta” product. While we have overcome a vast array of data issues and local “gotchas” over the last six months, we still have a backlog of hundreds of known issues and surely hundreds more yet to be discovered.
We seek to cover every community in the United States. Doing that will most certainly require a community effort. We need your help. If you see inaccurate or outdated data or have ideas on how we can improve our product or coverage, please click the “Feedback” link you’ll see hovering on the right side of any page on Niche Local, or contact us.
We’ll soon be releasing our Neighborhood Survey where you can share your opinions, reviews, and insights on your local area. Insider reviews, survey polls, and rankings like you’ll find on our Niche Colleges site are on the way, but for now, we’re very proud to be releasing Niche Local “Beta” into the wild.
The Enormity of the Task
The United States covers nearly 3.5 million square miles of land area. Within that vast expanse, the U.S. population currently sits at just over 317 million people, each with his or her own unique concept of “home.” Depending on the context, any one individual may relate to a state, metro, city, town, township, borough, village, community, neighborhood, block, or a single address as “home.”
Our goal with Niche Local was to provide 100 percent coverage of all places in the United States. While that has proven to be a nearly impossible task, mainly because of vast areas of the country that are very sparsely populated and/or unincorporated, we’ve been able to achieve almost universal coverage. This includes:
- 52 states (includes District of Columbia and Puerto Rico)
- 80 metro areas*
- 3,221 counties
- 55,465 cities, towns, and townships**
- 12,287 neighborhoods
- 33,120 ZIP codes
Curious about your hometown? Search for your hometown or ZIP code.
Same name, same place
Did you know that Washington, D.C., is represented as a state (District of Columbia), a county (District of Columbia County), and a city (Washington) with identical geographic borders and nearly identical data? Same thing for Philadelphia, PA—it’s both a city and a county.
Same name, different place
There are 295 places in the United States named “Washington” (e.g., Washington, PA) and 89 more with “Washington” in the name (e.g., Washington Heights). Not to mention, we’re already covering 39 colleges (e.g., Washington College) and 600 K-12 schools (e.g., Washington High School) with “Washington” in the name. As you can imagine, that can cause some problems.
Due to duplicate profiles and similar names, building an intelligent site-wide autocomplete search is quite a challenge and is something we continue to work on. A location’s size, proximity, and popularity can all act as competing factors in determining relevancy.
Data Sets vs. Colloquialisms
Everyday people don’t think like a U.S. Census data set. For the Census to have complete coverage, they had to create some towns (CCDs) where no such legal body existed. The majority of these fictitious towns are election districts. We removed them because it was determined a person would not define them as his or her hometown. Louisana, Maryland, Mississippi, Tennessee, Virginia, and West Virginia do not have any CCDs on Niche Local for this reason.
Further, Census data is not reported at the neighborhood level. Rather it is reported as Census tracts, blocks, and block groups, which generally do not directly correspond to real neighborhoods in everyday use. Nearly all data sets are also missing concept of “regions” that are referred to in everyday use—things like Long Island, the Outer Banks, and the Florida Keys.
While we’ve overcome some of these issues by mapping data to more “everyday” places, we still have a lot of work to do.
Data Holes & Inconsistencies
As mentioned above, the Census does not report data at a neighborhood level. To convert Census data into neighborhood data, we looked at each block group that had some land coverage within a neighborhood and added that block group’s data multiplied by the percent that the census block group overlaps within a neighborhood. For purposes of calculating the data, we assumed that the census block group is evenly distributed (though we know this is not necessarily the case).
We also ran into inconsistencies and anomalies related to reporting differences across states and local areas. For example, Louisiana uses parishes as the legal body of local government, and Alaska uses boroughs rather than counties. These types of anomalies create issues with when trying to aggregate and compare data across regions. Data holes also exist when reported data doesn’t meet the data sources’ standards for data collection. For example, Chicago’s method for reporting crime data does not match the FBI’s methodology, so some crime data is missing for Chicago on our site.
In addition, our weather and crime data is reported at the town level, so it was either extrapolated or removed from our other place levels.
More Challenges to Come
Although the public beta of Niche Local launches as a fairly basic product, it took a monumental effort to compile, process, and clean the data necessary to build it. Now that we have the foundation, we’re going to be adding user-generated content via open-ended reviews, surveys, and polls and developing rankings of places across the country.
Our goal for Niche Local is to make choosing where to live, visit, or raise a family a much more transparent and informed process. This public beta is a great first step, but it is merely the tip of the iceberg.
A few notes on the above numbers:
* The Census lists more than 80 metros, but they start to lose meaning as you get smaller. We went with the largest metro in each state plus the next 28 largest nationwide.
** This includes 27,700 CCDs or “county subdivisions” (like a township) and 28,299 CDPs or “census designated places” (like a town or city).