Admittedly the name isn’t very inventive, in fact, I don’t really know what to call them other than piles! The name aside I really love this concept, the idea that data can be displayed in physically accurate piles or mounds like this to me makes it a lot easier to parse.

Before I explain what the image is showing it’s best to discuss the concept behind and rationale behind the idea. Here is my twitter thread doing just that…

So using the above concept I applied it to a familiar dataset, I have been working with data from The Bus Open Data Service (BODS) for the past year so wanted to visualise how many unique buses were present in the system at the moment. Each sphere represents a unique bus on a unique journey per Local Authority Area (LPAs). The mounds or piles indicate the volume of buses in these LPAs generated around a centre point. Large piles/mounds = more buses. The concept itself is pretty simple and inspiration was taken from this wonderful visualisation for ‘Drowning in Plastic’.

 Following on from this concept with BODS I also applied it to bus occupancy data for Newcastle. The below visualisation shows data for the net change in people on or off a bus from 6am to 10pm on August 25th. Each sphere represents one person so the distribution and pattern of where people get on or off a bus is quite clear to see.

I think the concept is interesting and works in 3d, the mounds are easy to distinguish and therefore the pattern of activity on buses is easy to decipher. There is the usual 3d issue with occlusion and obviously being able to interact and pan around the scene would be hugely beneficial.

I came to this concept after struggling a little with a previous concept for visualising occupancy data. Prior to this, I was developing a technique that uses spacing algorithms to distribute spheres of geometry. The algorithm took place on a flat plane meaning that when the spheres were spaced out there was an inevitable ‘offset’ from where the sphere originated. Patterns were clearly visible but the concept maybe created a misguided view on where the patterns were taking place - it worked on an abstract level but any detailed analysis wasn’t advised. The visualisation styles I am referring to are below.

Whether the concept is piles or the flat distributions we see above both techniques have draw backs but I think what is useful is the snapshot view you can make of the general pattern of distribution for occupancy data. My research continues…