UCB — Fall 2013 / Week 7

Data in the wild

So far in class, our data has mostly come nicely formatted. But in the wild it’s rarely that nice. Today we’ll do one small exercise together and then work on scenario where you get data in a terrible, terrible format.


• Your pitches and your homework for tomorrow.

• A sample of R code and questions about last class’s exercise. Challenges/excitements.

• Office hours this weekend at 6

• Yet more free help on the wiki

• Possibly handy guide to data transformations


Sam and Nausheen are discussing a recent UGC map by the Washington Post this week.


You’re a data journalist working at the Chicago Tribune’s tribapps team. The city is generally good about making its data public – there’s a data portal that keeps data of every crime, and another news site, RedEye, maintains an app dedicated to tracking homicides in the city. Things were not so good last year – the city had its highest number of homicides in 5 years, which attracted national attention.

In this exercise, you’ve obtained data about successful gun traces in Chicago in the last 10 years. (A gun trace is where they find out where a gun found in a crime was purchased.) Go ahead and download it.

Each row in this data represents a county in the United States and how many Chicago guns were purchased there.

We’ll break into 4 (6 if you count missing people) groups for this.

Group 1: Erik Reyna, Sean Greene (gone for ONA)

Group 2: Jess, Angie, Chelsi (gone for ONA and/or sick)

Group 3: Sam Rolens, Aaron Mendelson, Samantha Masunaga

Group 4: Julie Brown, Diego Barido, Stephen Fisher

Group 5: Mihir Zaveri, Nausheen Husain

Group 6: Pearly Tan, Tawanda Kanhema

We’ll start up together, ask some questions, do some light formatting, then we’re off to the races.

Here’s your rough schedule

Starting at 4:45pm

State County

Starting at 5:00pm

Starting at 5:15pm

Starting at 5:45pm

Starting at 6:45pm

Starting at 7:00 (to 7:15?)pm


Same as before – post your sketches online with a link to reproducible R code. In this case, though, make a valid HTML page with a headline, readout, and some body copy (presumably answers to your data questions), and some questions you’d like to pursue if you have time. We’ll post a template or two that might be useful for you, but feel free to make your own, too.