Now I want to focus on what kind of information you can visualize with a treemap. So first of all, as you have seen, area is one of the most important visual properties of a TreeMap. With area, you can convey information about quantities. In general, what else you can visualize with a TreeMap other than some quantitative attributes with area. You can also use color. So typically you can use area and color at the same time to map two pieces of information at the same time. Now with color you can visualize both quantitative, ordinal, and categorical attributes. You can also have of course the hierarchy, the nesting sequence that gives information about containment and hierarchy. So here is an example very simple example of dataset showing you information coming from a company, it's a sales dataset. Here what I've done is to create a visual representation with a treemap, where every single rectangle represents a product category and this product categories are nested in a number of hierarchies, and the size of the rectangles represent the total amount of sales within each category and the color represents the average profit. So that's what is really nice about treemaps because you can see the hierarchy, you can see our quantity, but you can also see another quantity at the same time. So it's easy to spot in this treemap what are the product categories with the highest number of sales, the biggest rectangles but it's also easy to spot the rectangles that represent the higher amount of profit or the lowest amount of profit, in this case, we have only one red rectangle and it's pretty big so you almost can't avoid noticing it. So this is what happens very often with treemaps. There are areas that you just can't stop noticing because they stand out from the rest. So here is exactly the same dataset but now with an additional level of the hierarchy. We're going at the level of products. So every single rectangle here is one single product. If you notice, let me go back to the previous representation, is exactly the same areas and exactly the same splits. The only difference is that within each of these rectangles now, we have smaller rectangles that represent individual products. Once again here we can see single products that have very high amount of sales, much more than the rest, these are the big rectangles that stand out but we can also see products that have high amounts of profit and low amount of profit. And this is also somewhat independent from the size of the rectangles. So it's very, very powerful. So let me conclude by saying a couple of interesting things about treemaps. The first thing that I want you to notice is that they are incredibly scalable. Especially when compared to other solutions. So one solution that is possible to use to visualize a large number of categories with associated values is typically bar charts, but if you try to visualize so many products in a bar charts, is just going to explode visually. So this is an example. You'd need a very very very small bar chart in order to visualize these very very large set of products, it just doesn't work. Another little thing that I want you to notice is that in treemaps we can also use color to represent categorical information, which is something you have already seen in the first treemaps that I've shown you I want to show you again. So this is a treemap that shows a very large number of elements, in this case, this is information about the code coming from the Linux Kernel of a few years ago. So, every single rectangle here is one specific file, there are many, many, many files but in this case, what I think it's interesting to notice is that color is no longer used to represent some quantitative information but is used to represent some categorical information. In this case, we have five types. So for instance source files are yellow, header files are pink, text files are dark blue, and so on. This way you can see how different types of files distributing different elements of the hierarchy. So both kind of encodings are really useful in order to cover different kind of needs and visualization tasks.
