Making Sense of Census Data With Google Earth

mikemuch writes "Imran Haque has developed a mashup of Google Earth with data from the U.S. Census Bureau, called gCensus. The app uses the XML format known as KML (Keyhole Markup Language), which can create shapes and colors on the maps displayed by GE. Haque had to build custom code libraries (which he's made available as open source) that could generate KML for the project. He also had to extract the relevant data from the highly counter-intuitive Census Bureau files and store them in a database that could handle geographic data. gCensus lets you do stuff like create colorful overlays on maps showing population ages, race, and family size distributions."

Comments

Exciting!

[Prysorra]

Imagine what applications you cook up with this .....

Perhaps there's a way to fuse the presentation possibilities with Gapminder?

Re:Exciting!

[the_greywolf]

Incidentally, I worked on a project that used the same data to a very interesting end: (Fortunately, my NDA has expired by now.)

In our project, we needed to know exactly where and what coverage was needed on which FM channels where we could get away with putting in an FM translator station. We imported the census data in Access (which was surprisingly straight-forward), and used the Manifold System initially - and later MapInfo - to visualize the areas we could sneak in without interfering with existing (and future) stations.

We developed several Access apps that allowed us to build a list of new sites to set up translator stations: Over 17,000, if memory serves. In the March filing of 2003, a couple Perl scripts pushed 4,221 applications to the FCC, nearly all of which were automatically generated.

Over the course of the next 5 months, we went over each application by hand (again, using our custom software, combined with software originally designed for this purpose), while a couple of us developed our own series of scripts and programs to perform interference studies in MapInfo.

In the August filing, we withdrew several applications and improved nearly 2,000 others.

I left in March 2004, but last I heard, they had nearly 2,000 construction permits, and had actually built 5 or 6 of them. A few of the other permits have been sold for upwards of $500,000.

All in all, I can't say I really approve of their methods, but I can't really say I'm not proud of what we achieved, either.

not even m$

[mastershake_phd]

The Census Bureau has meticulously documented its data files--in a 635 page PDF file.

Wow, now thats a file format.

For the people, by the people.

[skoaldipper]

The Census is equally important as voting. Special interest groups representing minority organizations work closely with state and local governments when they draw up political districts. What an awesome tool to hold those officials accountable and give other groups a voice - open access for everyone.

Re:Part of this has already been done

[HarmlessScenery]

"A lot of this has already been done, although the site hasn't updated since google changed their API's: www.gcensus.com."

He mentions gcensus.com in the article - it uses Google maps rather then Google earth - and the API for Google earth apparantly allows more to be done with the visualisation of the data.

Others

[adickerson0]

I tend to use the Free AnalyGIS mash-up for Google Maps http://demo.analygis.com/google/default.htm For those who don't know much about AnalgGIS http://www.analygis.com/ I suggest checking out their web site. It is a pay service but the best of the bunch IMHO.

Talk about public service...

[jthill]

The real goal of the project is to democratize information by making data (such as political and environmental data) that's currently publicly accessible in name only, truly accessible to the people.

I'd like to see maps of the disparities between exit-poll and actual vote tally numbers, one map per election. This will make it possible, and not just "possible": once someone has putatively done the work, it'll be easy to check, because the raw data are available from trustworthy sources (cue cynicism in 3) so anyone can redo the map to check for distortions.

This makes whole classes of questions easier for mere mortals to answer, and simultaneously makes their answers easier for mere mortals to understand. It's huge.

Re:I knew from the headline

[Reality Master 101]

The word that conveys the work and technical details involved is "integration". Yes, it's four syllables.

(shaking fist) DAMN YOU KIDS! LEAVE MY DICTIONARY ALONE! I'M TOO OLD TO LEARN NEW LANGUAGE USAGE!

Or to put it another way, DAMN YOU KIDS! STAY OFF OF MY LAWN!

(It actually irritated me too, until I realized that my irritation was a symptom of my "over 30" age, and then promptly got over it)

Re:I knew from the headline

[Otto]

When I go to work, I don't make something meant to absorb butter and gravy.

You might, if you worked in Idaho.

Re:Down

[ihaque]

That's the wrong site. My site is http://gecensus.stanford.edu/, not gcensus.com.

"Catch gerrymandering"??? Here's a better way...

[Randym]

...such as using easily accessible GIS to examine voting patterns and election districts to catch gerrymandering and, potentially, election fraud.

You do realize, don't you, that *every time the political maps are redrawn* (i.e. every ten years) that gerrymandering is extensively used by every local state legislature? Gerrymandering is, after all, the process of redrawing the Districts so as to maximize partisan advantage. You don't need this tool to catch gerrymandering -- it's ubiquitous!

Rant aside, this app could certainly be a useful tool. The ideal -- nonpartisan -- political map would be drawn in such a way as to have the *sum of the sides* of *all* the districts to be a *minimum* while having the population of *each* district be within 1% of the average District population size for that state. (The Supreme Court has held that *some* variance in the population of districts is OK; I think that >1%, though, is *too much*.)

For example, say a state has 6.5 million people and 10 Congressional districts. Then each District must contain 650,000 +/- 1% (i.e. 1% here equals 6,500 people) and the sum of the sides of all the districts together must be a minimum. This leads to roundish districts and no possibility of gerrymandering (which, because of the torturous way districts are drawn, tends to *maximize* the length of the sides of districts).

The 'drawback' of this method, of course, is that only population -- and not historical voting patterns -- is taken into account, thus making it impossible to ensure that all the Democrats or Republicans or minorities are concentrated into just a few districts, as is done now by partisan legislatures. On the plus side, this would make more Congressional districts *unsafe* -- Congresspeople would actually have to get out there and *earn* their seats, instead of just sitting back and taking it for granted that their particular seat is safe because of the way that the Districts are drawn.

However, this scheme is unimplementable at the present time, due to the recent reauthorization of the Voting Rights Act (for the next 25 years), which actually *requires* gerrymandering when it comes to creating so-called "minority majority" Districts. (This ensures that minorities have adequate representation in Congress, rather than having their power diluted by being split among a number of other districts.)

For example, here in Michigan, blacks constitute about 12% of the population, but are highly concentrated in Detroit. If the above Voting Rights Act stricture was *not* in place, unscrupulous politicians could redraw the Congressional Districts in such a way (for example as long thin areas that had one end in Detroit and the other end outstate) as to ensure that blacks had *no* majority districts, and were a minority in *every* District that they were in -- clearly a violation of the Equal Protection clause of the Constitution.

So my logical plan for nonpartisan redistricting is -- unfortunately -- unlikely to come to fruition anytime soon -- for Congressional Districts anyway. However, since the case that was adjudicated by the Supreme Court *only* addressed issues of *Federal* redistricting, it might be possible for individual states to implement this plan as a way of making elections more contested and, hence, more democratic.