Tuesday, January 18, 2011

Some Contextual Charting

Context is a great way to lend meaning to a dataset.  One particularly fun way to contextualize is to display the same data set in various simultaneous forms, in an effort to cover some the information architecture bases.

The image above is a snapshot of a Visual Fusion demographics application showing a map of census tracts shaded by median income.  The top histogram plots what looks like a pretty-normal distribution of income across much of Michigan.  The histogram beneath that shows the distribution of the unemployment rate.  Hovering the mouse over any portion of the map or charts pulses the corresponding tract in the others concurrently.

Here are those same tracts shaded by relative unemployment rate.  The topmost scatterplot shows the relationship between unemployment rate and the vacant house count per tract.  The second scatterplot relates the unemployment rate to relative ruralness.  The area of the census tract is used as a proxy to characterize how rural a tract is, generally.  Larger tracts are assumed to be more rural, smaller tracts tend to be more urban.

So far we've seen examples of feed-level charting, i.e., charts that plot all the items that comprise a particular layer.  But what about charts for those individual items that comprise the layer (rather than a rollup chart where they all live together)?  That's certainly helpful as well.  One way is to make charts that are peers of each individual map element, so each shape on the map gets a distinct chart.  Here is an example of that, using a sparkline inside a tooltip; as the user mouses about, the snippet of information that appears includes a neat little stripped down line chart.

It is also entirely possible to use a simple chart as the map icon itself, so, when placed into a feed-level chart you would get a chart of charts.  But that's crazy!  And entirely possible.  But still pretty crazy.

No comments:

Post a Comment