Analyzing DC Bicycle Theft Data
June 1st, 2012 [programming]
I dipped into DC’s data catalog and extracted the bike theft data. They have five years of crime stats posted. However, the most recent year, 2011, has only 5 bike thefts reported. That’s after 629 in 2007, 729 in 2008, 551 in 2009, and 545 in 2010.
The crimes are recorded in the city’s database as crime = theft and method = bicycle. I’m baffled why this is recorded in a “method” field; this sounds like the crime was committed by someone on a bike. The city’s Crime Incidents page defines their usage thus:
Theft of another’s bicycle, as defined as a human-powered vehicle with wheels designed to transport, by pedaling, one or more persons seated on one or more saddle seats on its frame; or with any attachment to the vehicle designed to transport by pedaling when the vehicle is used on a public roadway, public bicycle path or other public right-of-way. The term “Bicycle” also includes a “tricycle,” which is a 3-wheeled human-powered vehicle designed for use as a toy by a single child under 6 years of age, the seat of which is no more than 2 feet from ground level.
2007 is the only year for which the data includes a “narrative” field. For bike thefts the narrative was typically a variation of the following samples:
- C1 REPORTS VARIOUS PROPERTY WAS STOLEN BY UNKNOWN PERSON(S)
- C1 STATES UNKNOWN SUSPECT(S) STOLE VARIOUS PROPERTY FROM UNDER HER FRONT PORCH
- C1 REPORTS UNKNOWNNOWN PERSON(S) CUT THE LOCK TO HIS BIKE AND STOLE SAME FROM HIS GARAGE
You would think a theft crime (and a burglary, which involves breaking and entering) would have a sub-table to list all items stolen. Instead, by listing bike thefts as a separate category of theft, it gets mixed up with tricycle thefts and burglaries where a bicycle might be only one of several items stolen.
The center of gravity has moved, as the image above illustrates. The average bike theft location in 2007 was at 15th & S NW. By 2010 it moved to Columbus Circle in front of Union Station.
You’ll see only three dots for 2011 because two thefts were recorded at the same location, and a fifth had no coordinates. (And I have no idea why the data shows only 5 bike thefts.)
The QuickTime movie on the right shows an animation of bike theft data, from 2007 to 2010. Each frame shows a year-long period, moving the starting date up by a week with each new frame.
I have no theory why the bulk of bike thefts migrated from Northwest to Capitol Hill.
Aside from the missing chunk of data from 2011, the data is obviously impacted by the large number of unreported crimes. I wonder how many people register their bike with the Metropolitan Police Department, or the National Bike Registry. (See Bicycle Registration in the District of Columbia.)
I was happily surprised to see that of the 2,454 thefts in the five-year period, only 10 were missing geographic coordinates.
Bike thefts are unsurprisingly seasonal, with 16% of all thefts in July, and only 3% in December. (Looking at all categories of crime, July is also the busiest month, but with only 9% of the total, while it’s February that’s the slowest month, with 6% of the total.)
The data is missing any information about the resolution of the crime. It would be nice to know whether the crime was solved, and to have information about any perpetrators.
My last look at DC crime data was Animated Map of Homicides in Washington, DC. For a different type of animated bike data, see Watching Bikeshare Stations Grow Unbalanced. The movie’s background image is a modified version of the Nighthtvision map from MapBox.