Now, I want to introduce two more methods to represent trees and hierarchies through space subdivision. So, these two techniques are the sunburst and the icicle plot. Let me start with the sunburst. So, the sunburst is a visualization technique that strikes a balance between the need of showing the hierarchical structure but at the same time, being as scalable as possible as the way treemaps are. How does it work? It's actually pretty intuitive. So, at the center of this circular visualization you have the root node. Now, around the root node you have some sort of doughnut, and the doughnut is split into as many areas as many children the root node has. The amount of space that is devoted to each one is proportional to the values that they represent, very similar to what we have with treemaps. So, that's the first layer. The second layer is exactly the same thing. For every single node coming from the first layer, it's split progressively into as many children as these node has. There are as many areas as the children, and their size is proportional to the values associated to them. So basically, starting from the root node in the center, this is extending in a radial direction from the center to the periphery. Another visual representation that is very similar to this one is basically the equivalent in the planar projection and it's called the icicle plot. It's exactly the same idea except that it doesn't use our radial layout. So, at the top, we have the root node. Below the root node, we have as many rectangles as children and their size is proportional to their values. Then, the next layer and next layer, next layer, and so on. So, very intuitive and very interesting in terms of striking a balance between trying to scale as much as possible and at the same time, showing the structure. So, the sunburst visualization has been proposed originally in a paper that is called an evaluation of space-filling information visualizations for depicting hierarchical structures. I think what is interesting in this paper is that the authors tried to evaluate whether the sunburst has any advantages compared to the treemap visualization. They found that it does actually perform better than the treemap in a number of tasks. It's important though to keep in mind that in this study, I think when the study was conducted back then, this squarified treemap visualization or solution was not popular yet. So, the authors compared the sunburst visualization with the treemap version using the slice and dice algorithm as we have seen as a number of problems. The icicle plot has also been introduced in a paper much earlier in the 1980's and it's been developed mostly as a way to represent results coming from hierarchical clustering which is the technique that I presented earlier, and it's very popular in domains like biology and comparing in general similarity across different objects.
