Before class
Reading
- "Capitol Assets" by David S. Fallis, Scott Higham and Kimberly Kindy, The Washington Post, 2012. Presented by Cecile Schilis-Gallego.
- Fast & Furious", by Isaac Lee, Daniel Coronell, Gerardo Reyes, et al., Univision, 2012. Presented by John Ismay.
In class
- Story reviews
- The in-class presentation on visualization
Resources
Using visualization as a reporting tool
Visualization is an important part of your ability to see stories in your data. For this stage, you want to try a lot of tools and a lot of views of your data, worrying less about the aesthetics and more about the revelations you might find.
- A chapter written by Sarah for the Data Journalism Handbook that never seemed to appear in the final version
- The draft re-write of the visualization as a reporting tool chapter from Numbers in the Newsroom
- Nathan Yau has a great website called FlowingData of examples that help you see what is a useful and not-so-useful visualization. His book, Visualize This: The FlowingData Guide to Design, Visualization and Statistics has some great tutorials.
- Edward Tufte may be the father of modern thinking on visualizing data. His book, "The Visual Display of Quantitative Information" might be found on the bookshelves of almost every graphic artist.
Tools for visualization (we'll hit mapping and GIS later in the semester)
- Most people swear by R's graphics -- the single best reason to learn the language. GGPLOT2, created by Hadley Wickham, is a versatile way to create charts from almost any form of data. There are plug-ins that can turn them into interactive graphics.
- Tableau and Tableau Public are supposed to come to Macs in the "2nd quarter" of this year. There are mixed feelings about
whether it's worth the learning curve, but it's a little less daunting than starting in on programming. Your IRE membership
entitles you to a free copy of the desktop version, which usually costs about $2,000. Here are some
handouts from IRE to get you started. (You have to be logged into IRE to get them.)
Just a warning: once you have created your visualization, it's hard to share the desktop version with your colleagues. The Public version is public to everyone. So it may not be the right answer for a lot of projects. - While Excel's graphics are, well, unhelpful, it has one feature that can be very useful: an implementation of Tufte's sparklines -- small, word-size graphics that help you distinguish patterns from a chart of numbers.
- Google Charts and their cousin, Google Fusion Tables, can help you create quick charts from small, simple statistical summaries. This usually isn't that useful during the reporting stage, but can be easier than some of the alternatives. Getting your data into shape means creating a JSON file from your csv or Excel sheet. Try using Shan Carter's Mr. Data Converter for that chore.
- D3 is a Javascript library created by Mike Bostock, now of the New York Times. It's the successor to Protovis, which was created in the Stanford Visualization Lab. Although intimidating, if you're good at Javascript and can follow examples, it's probably the most versatile visualization tool.
- Simile Exhibit is a project out of MIT. It's good for simple maps and interactive timelines, especially when you want to filter small datasets at the same time. You may need Mr. Data Converter for this, too.
- The missing link in exploratory visualization is a way to view timelines and calendars for chronologies. We took a stab at this at Duke University, enlisting the visualization geniuses Martin Wattenberg and Fernanda Viegas to create a prototype tool called TimeFlow. It's rough -- make sure to keep backups of your data in Excel -- but it can be useful for exploration