Monday, April 14, 2014

Volume and proportion

A question posted on the International Association of Crime Analysis forum last week caught my eye. It was by a detective at a municipal police department in Texas, who was trying to figure out how to create a map which would depict the relative number of thefts, burglaries, and robberies in each of his city's three police districts.

Since he wanted to do this using ArcGIS (something I'm pretty comfortable with), I replied to the list with my suggested method. I suspect crime analysts who lurk on the IACA list are somewhat shocked when a person in executive management responds with a solution to a technical GIS issue. I'm probably older than most of their fathers, too, but from time to time I like to remind myself that I can still do things like create pivot tables, hammer out a little html code, and wrangle a GIS project to make it do what I want it to do.

At any rate, this is similar to what he was trying to accomplish. It is a map of robberies in Lincoln from 2009 through 2013, depicted as five pies, one for each of the Lincoln Police Department's command areas. The differing size of the pies reflects the relative number of robberies within each Team, while two slices of each pie are business robberies and non-business robberies.

It's a simple graphic that at a glance conveys information about both volume and proportion. As you can see, the Southwest Team has a much larger volume of robberies, and the non-business slice is way bigger than most of the other teams. Basically, street robberies are the issue in the Southwest and Center Teams, while business robberies (though still fewer than non-business) are more prevalent in the Northwest, Northeast, and Southeast teams.

Click image to enlarge
For analysts wishing to create this effect with ArcGIS, take your polygon layer (in my case, the five police teams) and add fields for the pie slices and their values (in this example, business robberies and non-business robberies). Under "Properties" change the symbology to "Charts" and select a pie chart. A stacked bar chart would also work well with these data. There are other settings for various options, but you'll get the drift.


Anonymous said...

Just based on the pie charts it appears the ratio of business to non-business for NW, NE and SE teams are about the same while SW and Center have about the same ratio. SW and Center obviously have a much higher population density, but would that be a reason for the difference in the ratio?

Tom Casady said...


Well, this post is really just about a visualization technique for GIS data--not really an analysis of robbery--but you're questions are good. The differences in proportion could be reflecting such things as population demographics, or where in Lincoln businesses tend to be located, for example.

Mark said...

Pivot tables, dang now I am imprssed. Next you'll talk about using VisiCalc?