Data Clusters
Current developments in Artificial Intelligence technology yield dividends for understanding how data is used in visualization. Data wants to be visualized. To that end computing expands conventions of graphing that give form to data. Shape is meaning is concept. Like geometry, seeing is believing, that is, proof.
The Bell Curve familiar to everyone is but a slice of the data pie chart. Imagine the Bell Curve as a profile view of a literally bell-shaped object having three dimensions or an x, y, and z axis. In engineering design such a “slice” is called a Section view.
Viewed from above through the y axis, the Bell Curve appears to be concentric circles -like a bullseye. The center circle, or “eye” of the bullseye, defines the mean distribution of data. The outer circles define standard deviations.
If the combination of section view and overhead view are viewed through the z axis the body of data becomes a 3-dimensional object. Again, like a literal (brass) bell rocking and ringing on its pinions, back and forth.
Freeze that rocking motion and consider a single position of the “bell” as if it were a transparent object hanging in space. All data now is observable, like looking at tropical fish in an aquarium. Instead of a simple sine wave (section), or concentric circles (overhead) plot, the data take the shape of cubes, cones, blobs, and yet more complex iterations.
Back to the engineering design analogy, a hanging chain, for example, doesn't conform to geometry but to Gravity. It's not a parabola (not exactly), but a Catenary curve. Its path is defined by Calculus, not Algebra. A discussion of physics is beyond the scope of this blog entry. The point is that material data describes reality in the real, physical, world.
An example which leaves something to the imagination is the mapping of social data. Consider a crime map of incidents in a given neighborhood. Real data are not distributed in a perfectly concentric circle (unlike the bullseye example), but are stochastic, dropped here and there in clusters, or isolated from the rest, completely absent in places, with a few right on the edges.
Next, correlate property value with reported incidents involving police data. What factors determine the physical location of cohorts? That “first month free” apartment rental might be a bad bet if there is a month's termination clause, and the renter wants out before (not after). Maps of data can be overlaid in superposition to powerful effect. A picture of the reality of the world of facts is composed by montage, stacking, intersection, and superposition.