Now, before we move on I want to introduce a very useful conceptual model that helps you understand at what stages of data transformation, interaction can play a role. What you see here is called information visualization pipeline and there are a number of stages. So, we start from data. Data typically is transformed in order to shape it in a way that one is able to extract the information that is needed for the visualization that we have in mind. The next step is to transform the data, the transformed data, into a visual abstraction. What is a visual abstraction? It's basically a collection of graphical objects or a description of how to go from the actual attributes and data items that you have in a table, to the graphical properties of the objects. So for instance, what kind of visual graphical objects do you have? What is their position on the screen? What is their color? What is the shape, if they have any and so on? So the logic that transforms data into something that has a graphical format. Then we have one last stage that is rendering which basically transforms these graphical abstraction or visual abstraction into the actual rendering on the screen of your graphical device. Now, what is interesting is that at each of these stages there is a transformation. So, in the first stage there is data transformation, in the second stage there is visual mapping, so the idea of deciding how data attributes map to visual attributes. So for instance, in the previous examples, the example that I showed you in the previous videos, we have that the graphical obstruction is dots that are colored according to food type, and their position is in a scatter plot and it's derived from two attributes which are the amount of carbohydrates and the amount of calories, so this is the mapping. The last step is view manipulation. So once something is rendered on the screen, it can still be manipulated in order to change the perspective used to watch or observe this visualization. So now in order to make this more concrete, I want to give you another example that is still based on the same example that I've shown you before, but now I want to walk you through specific types of interactions that actually insist on these three different steps that I've just described of the information visualization pipeline. More specifically, I'm going to show you interactions that have effect on the data, interactions that have an effect on the visual mapping, and interactions that have an effect on the view. So, let me show you. So what you see here is a slightly different representation, I am building this interactive visualizations with Tableau Software, which is really powerful and allows you, other than creating visualizations it also allows you first of all to make it interactively and to create visualizations that are interactive. So, this is the scatter plot that I've shown you before with nutrition data. Every dot is one food product and then we have carbohydrates and calories. We have the food groups as before and we have an additional filter that is used to filter out data according to how much water the foods contain. So, if I move the slider you will see that the data on the scatter plot is updated. So, if I move the slider you will see that I'm removing all those products that have water below the value 35, 36.5. So effectively what is happening when I'm doing that is that I am filtering the data, then the data is filtered and it's visualized in a way that I can see which data items stay there. So, this is a type of interaction that has an effect on data, okay. Then since it has an effect on data it also has an effect on visualization. But it mainly intervenes at the data level. Now, let me show you an example of interaction that actually has an effect on the mapping. So, this scatter plot is already the effect of a specific kind of mapping, as I said before. I mapped amount of carbohydrates to the X-axis, the amount of calories to the Y-axis, and the food type to color. So let's say that now I want to change this visualization in a way that the size of the dots is proportional to the amount of water that is contained in these foods. So how do I do that? So I select water, I drag water to size, and now as you can see the size of the dots is proportional to the amount of water that these foods contain. So this is an example of an interaction that changes the mapping, this is not changing the data, it's changing the rule that is currently used to map data to visual abstraction. So now let me give you another example of the last type of change that you can make, a change in the view. So, let's say that I want to see more, in details, these food products that are here. They form a nice group, a nice cluster. So, one way to do that is to double-click here and I'm zooming in this area. So as you can see, the view is changing. In particular what is happening here is that I'm zooming in into a specific area. So this is not a change in the data and it's not a change in the mapping, it's a change of the actual view. So, what I want you to remember from this video is that interaction can affect different stages of the pipeline, and in particular these stages are transforming the data, creating a mapping between the data and the visual abstraction, and finally manipulating the view.
