- Following my post on Chart-Making: Cures for Loss Aversion, Andreas at More information per pixel shows how to group series using a diverging color pallet. Interesting discussion, too. Kelly, from Process Trends, join the conversation with a new blog, Charts & Graphs (congrats!). Derek, at Information Ocean, uses a log scale to show a hidden pattern. I like to use a log scale, but unfortunately it is difficult to explain it.
- Public beta release of ICharts. The you tube for charts? Again? That’s what Swivel said. I still prefer ManyEyes.
- Dataset of the day: Top 100 commercial banks with exposure to risky home loans. Not exactly my kind of visualization, but good datasets and a link to Finder!, a “browser-based application for finding, organizing and sharing GeoData in common formats”.
- Keynote Presentation by Stephen Few at Tableau Software. Terrible video recording and nothing really new, but nice sense of humor.
- NY Times: The Best and Worst of Data Visualization. The New York Times spoils us all with great information graphics, but some times something goes terribly wrong. Robert Kosara discusses some of those bad examples.
- Speaking of bad examples: JunkCharts adds another in Bubbles of the same size.
You may also be interested in:
I like to use a log scale, but unfortunately it is difficult to explain it.
xkcd helps out
Actually I was being really lazy using a log scale. As Jon Peltier had pointed out before, it’s not really appropriate for a 0-1 distribution, it should be used for a 0-infinity distribution. But it was easiest to just tick the “exponential” button in Excel.
See? Very difficult to explain…
Seriously: If you want a safe dataset to play with log scales merge the datasets in the link above. It tells a very interesting story of the home loans mess.