Before we conclude, let me summarize the main ideas that we discussed in this module. We started by discussing the idea of visual encoding and decoding. What is visual encoding? The rules, the methods that transform data into a graphical representation. What is visual decoding or graphical decoding? Is the opposite. You observe a visualization and you try to extract out of it the rules, the mappings that created the visualization out of data. Then more in details I introduced the idea that every visualization is made of two main graphical components: Marks and channels. The marks are the graphical elements that represent the objects, the data items. And the channels are the graphical properties that encode properties of the data. What we called data attributes. Then we said, okay now that we know how to encode and decode visualization and what the language is, how do we know whether a given encoding is good or not? And we introduced two main principles. The first one is the expressiveness principle and the second one is the effectiveness principle. The effectiveness principle is basically the idea that the visual representation shouldn't communicate information that is not actually contained in the data. The effectiveness principle is that you should try to use the most effective channels for the most important information that you want to present. And then we looked more into what it means for a channel to be effective. And I presented to you a table or diagram, that shows you two main parameters of effectiveness. One is that, regarding the fact that the channel needs to be appropriate for the kind of information that you are presenting, in particular you have channels that are appropriate for quantitative and ordered attributes, and you have channels that are appropriate for categorical information, categorical attributes. Then I also presented the ranking in terms of accuracy. So we saw that some channels are more effective than others, more accurate than others at presenting quantitative information, and some other channels are more or less effective at presenting categorical information. Finally, we went through a few exercises showing you how these two principles and the ability to identify graphical elements and encoding rules is useful for evaluation and design of visualization. So it's a very very powerful tool. With these tools, you can evaluate existing visualizations and you can design new visualizations or redesign existing visualizations. Very important tool. Finally, I concluded with introducing the idea that there are a number of contextual components. We don't have all only marks and channels, but typically we have a number of additional components that even if they are contextual, they are actually really really important and useful for interpreting and reading values out of a visualization. In particular, I talked about labels, legends and annotations as a way to help the reader interpret the meaning of the objects that are displayed on the graph. And then I talked about axis, grids and trend lines that help you comparing and reading values out of the visual representation.
