With the possibility of thousands of data points, how might we represent clusters of data in a way that doesn’t overwhelm people yet could still reveal top-line information about a dataset?
This is the idea of going from low-resolution to high-resolution. In this example I’m using clusters of clusters. On initial viewing each cluster is represented by simple points. The size of each point within each cluster is relative to the size of the more high res cube representation. Even in this low res view certain details about that dataset might be able to be ascertained. These low res points in effect create simple patterns.
The user can then zoom in on and bring to the centre a certain dataset which then has the effect of revealing a more high res view of that dataset, hopefully revealing more about the data contained within. This could then lead to a way to then expand that dataset to fully explore it.
I liken this action to that of tuning in a radio, quickly scrubbing the dial back and forth to try and find something of interest, yet done visually. The use of sound (explored later) could also be added to further explore the data.