Log in

No account? Create an account

October 5th, 2012

fedora 蓝色小药丸

Tribune "Crime" app

The link to Chicago Tribune "Crime" app is http://crime.chicagotribune.com/chicago/

A side notice: I got the link from a great meetup event "DataPotluck" #4. The meetup event is in turn sent by David Eads, one of the developer of that app, to the mailing list for the "FreeGeek" group he founded. The "Crime" app is the main topic on that evening. Another developer Joseph Germuska, or Joe, and David made the presentation.

The app grabs data from government data portal, analysis and create pages. The scripts used to import those data can be found in Joe's GitHub repo, which also contains sample city map data and crime data. The scripts use PostgreSQL to import CSV data.

The page is really beautifully decorated. The front page has a huge tagline "Find out about crime on your block, in your community, along your commute, and more". Then a big table to the left listed all the 77 community areas of Chicago. On the right is a search box and a map. Clicking on any links goes to specific community area page.

The page for a community area contains much information -- it resembles a security factsheet I worked on. There is a "summary" part that consists of a map and a summary table. Clicking any summary item would change how that type of crime is displayed on the map. Crimes are divided to 3 types: violent, property, and quality-of-life. Combined map data and crime data, every crime case is marked on the map as a color dot. The map also conveniently show neighbourhood communities by checking a switch. The crime density, rank, and changes to the same 30-day window of last year are also included in summary.

The "In context" section shows some information of this area from US Census and community survey. A security factsheet would also include this kind of section. But the most resemblences come from the "breakdowns" and "historical trends" sections: it used the same multiple pie charts and line charts to describe crime details. Remember there are 3 types of crimes? That is the "asset allocation" or "sector allocation" of crime information. The "sector rotation" does worry oneself if read closely. A final section is "locations", where crimes actually happened. It is not the "portfolio holdings" table, because people would always avoid heavily weighted items.

To present city map data and crime data it used MapQuest, OpenStreetMap data, and Leaflet.js. On the page they are publicly thanked upon, or else I would have no idea (reading the starter's document now). I'm not familiar with GIS related technologies, but pgsql looks so good (the shell scripts uses it to import CSV map data and do some manipulation)

The application is an effort of opening up government data. Many people presented said they are part of the OpenCity initiative. The application does not say much about license of its analyzer code and analysed data.

There is an article for this application on cjr.org,[excerpted]
Germuska said the Chicago Tribune would not have been able to embark on this project without the city’s database. Crime data is volatile; it changes as crimes are reclassified. If this project was produced with scraped data, Germuska said he doesn’t think it would be as authoritative.

The public data used to build the Tribune’s new site runs on Socrata, a company that provides platforms for open government data. Founded in 2007, its clients include the cities of New York, San Francisco, Austin, and Seattle. Chicago uses Socrata to power its data portal,which holds over 200 data sets ranging from crime statistics to food inspections and public school progress reports.

The Tribune’s map updates as new crimes are reported. The app team would like automate this process so that when the data held in the Socrata data portal refreshes, it pumps the new figures into the crime app. Currently, that process is not updating correctly, so the app team has to monitor updates.

The CSV files are fairly large (hundreds of megabytes). How to decide what and how to present them is really a hard topic. I have to admire the beauty of this application. And the great intent behind those data and applications -- why these people care so much about open data and open government?

A similar application is SpotCrime website. User can choose to subscribe crimes within specified distance as daily email.

Read more...Collapse )