Wednesday, January 20, 2010

Walkability Map
























Warning: no baby updates. In my small amount of spare time recently, I created this map measuring how walkable different parts of our city of Shrewsbury are. It turned out to be somewhat less walkable then I expected, mostly because there is only one private school in Shrewsbury and because the train tracks make it longer to walk to the Dierberg's retail complex.
This was inspired by the trademarked Walk Score (http://www.walkscore.com/) developed by FrontSeat. I looked through the methodology on their website and found some room for improvement.This works by finding places that most people travel to (stores, schools, restaurants, etc.) and measuring the distance between your home and the destination. The more places within walking distance of your house, the higher your Walk Score is.

The results, which are pulled from an online yellowpages, aren't all accurate. For example, it classified the seminary next to our house as a school and a restaurant which may be technically correct but probably doesn't matter to most people. Also, each destination is given the same weight, you get just as many points for being near a coffeehouse as you do for being near a school or department store. The other big issue is that the Walk Score measures the distance as a straight line, regardless of how you actually travel to get there. So if streets don't connect, or if there is a fence or lake between you and the destination, it expects you to walk through it.
I tried to improve this methodology by concentrating on a small, familiar area which allowed me to verify all of the destinations. Also, I created my own scoring system which gives more weight to places that (I think) people care more about being close to. In addition, to take into consideration how walkable an area is, not just the distance, I took into consideration whether or not each street had sidewalks. For streets with no sidewalks, (determined from aerial imagery) I calculated an extra 30% distance that you have to walk because people are less likely to walk that route if there are no sidewalks.










My final improvement was to use distance traveled over the street network instead of straight-line distance. GIS software allows you to create areas representing the distance from a point over a network.



















Walkability Point System: 1/4 mile or less (1/4 to 3/4 mile)
Total Possible Points = 100
School: 14 (7)
Grocery: 22 (11)
Park: 10 (5)
Department store: 16 (8)
Bus stop: 4 (2)
MetroLink station: 14 (7)
Restaurant: 16 (8)
Bar: 4 (2)

I then added laid the distance to each destination over a map of all properties in Shrewsbury. Each property which overlapped with a distance-to-destination shape (see above pic) recieved points for that category. I then added up all the points for each property and colored them according to their score.

1 comment:

Aaron O. said...

Impressive work. It looks like you've spent some time looking at ways to improve the Walk Score algorithm. I've been doing some similar work over at Walkshed.org. I'd love to hear what you think.

Again, great work.