Ask HN: Is Geospatial MultiPolygon Data interesting to anyone?

#1

Post by iamjpg » Thu, Mar 17, 2011, 9:18 PM UTC

So about a year ago a friend of mine and I succeeded in accumulating Geospatial MultiPolygon data for states, counties, cities, zipcodes, boroughs, and neighborhoods across the US. After getting the data loaded into a PostGIS enabled Postgres database we both got busy and kind of put our little pet project on the back burner.

We're currently revisiting the project and working on finishing up the API. Yesterday a thought occurred to us that we couldn't answer: Will this data be interesting or useful to anyone other than ourselves?

We would love to get the thoughts of the HN crowd figuring that if the data isn't interesting to anyone here it might not be interesting to anyone :)

You can see the data in action here:

http://geolayr.com

Thanks for the feedback!

Re: Ask HN: Is Geospatial MultiPolygon Data interesting to anyone?

#2

Post by tgrass » Thu, Mar 17, 2011, 9:24 PM UTC

Very, very interested. Please keep me posted (sent you an email).

Re: Ask HN: Is Geospatial MultiPolygon Data interesting to anyone?

#3

Post by iamjpg » Thu, Mar 17, 2011, 10:17 PM UTC

tgrass wrote: post #2

Very, very interested. Please keep me posted (sent you an email).

Received your email and will be in touch as soon as the API is finished.

Re: Ask HN: Is Geospatial MultiPolygon Data interesting to anyone?

#4

Post by epc » Fri, Mar 18, 2011, 2:07 AM UTC

What's the source for your ZIP code data? The data for 11201 is very wrong (11201 is in Brooklyn, geolayr displays it as Governor's Island, Ellis & Liberty Islands).

Re: Ask HN: Is Geospatial MultiPolygon Data interesting to anyone?

#5

Post by iamjpg » Fri, Mar 18, 2011, 3:36 PM UTC

epc wrote: post #4

What's the source for your ZIP code data? The data for 11201 is very wrong (11201 is in Brooklyn, geolayr displays it as Governor's Island, Ellis & Liberty Islands).

It's Census data - unfortunately from 2001 - we're preparing to upgrade with the latest data from the 2010 Census which is just now becoming available...slowly...

Thanks for the feedback on 11201. I anticipate issues like this in the data set.

Re: Ask HN: Is Geospatial MultiPolygon Data interesting to anyone?

#6

Post by epc » Fri, Mar 18, 2011, 3:51 PM UTC

iamjpg wrote: post #5

epc wrote: post #4

What's the source for your ZIP code data? The data for 11201 is very wrong (11201 is in Brooklyn, geolayr displays it as Governor's Island, Ellis & Liberty Islands).

It's Census data - unfortunately from 2001 - we're preparing to upgrade with the latest data from the 2010 Census which is just now becoming available...slowly... Thanks for the feedback on 11201. I anticipate issues like this in the data set.

May want to verify how you're parsing the data set, am finding several ZIP codes which are close, but not quite. For example 12590 is drawn accurately (as I recall, haven't lived there in a decade), but when I enter 12603 I get the same map for 12590, when 12603 should be farther north as it's poughkeepsie. I tried this in two separate browsers so it wasn't a caching thing unless you're caching by client IP.

Another example: 11202 shouldn't have a map at all, it's reserved for the central PO in Brooklyn (Cadman Plaza GPO), instead it's being drawn as a region in North Brooklyn.

Another example from outside Brooklyn: 60515 (Downers Grove, IL) looks accurate, but 60516 (southern Downers Grove) appears to be drawn using the 60517 (Woodridge, IL) boundaries, and when I punch in 60517 it gets drawn using the boundaries for 60561 (Darien, IL).

I haven't played with GIS data for several years but remember than there was a set of ZIP code data sets around 2005 which had accurate latitude/longitude points but the ZIP codes were frequently off by one (I ended up having to manually massage the data because I couldn't find a pattern to the way the ZIP code was appearing in the data).