Walk Score Neighborhood Ranking Methodology
We sampled the Walk Score of 1,123,855 locations in the largest 40 U.S. cities to create walkability heat maps and rank 2,508 neighborhoods.
Data Sources
Walk Score algorithm: The Walk Score algorithm uses the Google Local Search API to retrieve nearby amenities (business listings, parks, etc.). Read the known issues with the algorithm.
City boundaries: We use the U.S. Census 2000 "Incorporated Places" for our city boundaries. Note: this is the city boundary not the entire metro area.
Neighborhood list: We use Zillow's list of neighborhoods to define the neighborhoods in each city and their shapes. If a neighborhood intersects a city boundary, we include it in our rankings. We only include neighborhoods with more than 500 people. There is no Zillow neighborhood data available for Oklahoma and Indianapolis.
Population data: To weight Walk Scores by population density, we use the U.S. Census 2000 SF1 Census Block Demographics.
Largest 40 U.S. cities: We use Wikipedia's list of the largest U.S. cities.
The Walk Score Point Grid and Population Density Weighted Walk Scores
The point grid.
Within a city or neighborhood, we sample the Walk Score of approximately each city block. To do this, we create a grid of points spaced roughly 500 feet apart (.0015 decimal degrees; exact distance will vary with latitude).
We weight the Walk Score of each point by population density so that the walkability rankings reflect where people live and so that neighborhoods/cities do not have lower Walk Scores because of parks, bodies of water, etc.
We use the following algorithm to calculate a population density-weighted Walk Score:
For each point in the grid:
- Expand each point by .00075 decimal degrees to create a grid cell
- Intersect the grid cell with the census blocks it intersects and for each census block:
- Calculate % of the census block the grid cell intersects
- Multiply that % by the total population of that census block
- Sum these partial populations to get the grid cell population
- Add the grid cell population to a variable called total_population
- Calculate the Walk Score at the center of the grid cell and multiply it by the grid cell population to get the weighted Walk Score
- Add the weighted Walk Score of this grid cell to a variable called weighted_walk_score
To calculate the Walk Score for an entire neighborhood/city, divide weighted_walk_score by total_population for the points within the boundary of the neighborhood/city.
The population total we display for neighborhoods/cities is the total_population variable mentioned above.
To calculate the Walk Score for a city, we include only points within the city boundary. This may exclude points that are inside a Zillow neighborhood boundary but outside the city boundary.
Generating the Walkability Heat Map and Neighborhood Rollovers
Before/after smoothing.
Heat maps: To generate the walkability heat map, we create a very small grid and assign each grid cell a color based on the Walk Scores of the surrounding points. A spectrum of red to green is used to represent the range of Walk Scores from 0 to 100. For visual clarity, we use the U.S. Census 2007 TIGER/Line data to remove water features greater than .0001 decimal degrees in area (about 3,500 sq. feet depending on latitude) from the heat maps.
Neighborhood rollovers: We show the Zillow neighborhood boundaries on the map when a user hovers over them with their mouse. To do this quickly, we simplify the Zillow neighborhood polygons using the Douglas-Peucker algorithm before transmitting them to the browser. In the browser, we use JavaScript to index all of the polygons using horizontal and vertical slices. This allows us to quickly determine which polygons might be intersected by a given mouse location before doing more detailed boundary checking.
Technology Stack
All geospatial calculations were performed using PostGIS. The heat maps are layered on top of Google Maps using MapServer and TileCache and served from Amazon EC2 instances. If you're interested in learning more, Paul Smith wrote an excellent article describing the open source mapping stack.
Walk Score Consulting
Wak Score offers custom mapping services. Contact us for more information.
View the Seattle Neighborhood Rankings.
Special thanks to Tom Blumer, Josh Livni, and the Walk Score advisory board.
GIS consulting provided by:
