Okay. Now, what we need to do next, is to discuss a little bit of what are the implications for design for the accuracy ranking, that we just saw. Okay, you can think of the ranking as roughly grouping the channels into three main groups. So, position is always the best channel and this is followed by length and angle and finally we have area. So these are channels that are very commonly used to represent him for quantitative information and more or less, roughly speaking you should always keep in mind these hierarchy, position is past length and angle are a little worse and area a little worse than length and angle. Okay? Keep in mind, that this is true when we are talking about the problem of visualizing a quantity and comparing quantities. But not all visualization tasks are about that as we will see later on. Okay. So the rule is, or the guideline more precisely is prioritize high ranking channels in your design if you can. Okay? And also related to that, do not expect your viewers to be able to make accurate comparisons or accurate estimates out of your visualization if you're using channels that come from the low ranking area. So, that's a general guideline that you may want to follow. Let me give you an example to make this more concrete. So in the following slides, I'm going to show you the same data represented with different charts and this is a simple data set. It's a data set describing sales coming from a private company. In this specific case we're using three variables, we have the profit amount by different categories of products and how it changes over time so the question is, how does the profit change over time for the three categories of products that we have in this company? One way to show this information, is to use what is called a stacked bar chart, which is exactly what you see here. So, on the x axis we have time, on the y axis we have the amount of profit, but each bar is also split into the three categories of products. Now, think about it when you're trying to compare the evolution over time of profit across these three categories, you are visually trying to compare the length of the bars and as we know, comparing lengths is not particularly effective especially compared to comparing positions. And that's why, trying to extract this information from this chart is not particularly easy, it's possible but it's not best. Here is another example, I'm presenting exactly the same information using a series of pie charts, each single pie chart corresponds to one specific time, sorry, month. The pie is split into three segments each one representing the amount for that category for that month. Once again, if you try to see what is the temporal evolution of the three categories and compared them, is not easy, not easy at all. And what are we using here? We're using an encoding method that uses angle and area, which again, they are not particularly effective. Here is once again another example, very similar to the previous one in this case we are using area, and area once again, is not particularly effective. Here, in this bar chart, we are using position. So the position of the top end of each bar is communicating information about the quantity for each one and if you observe this chart for a moment, you will see that extracting information the temporal evolution over time across the three categories it somewhat easier than before. Once you switch to a line chart, that uses position once again, an angle to represent information about variation, this is even better and I would say that's probably the best chart, if the goal, the specific task is to see the evolution over time and how the different categories compare one to another. Now, one thing to keep in mind, is that you should consider the ranking of visual variables or channels as a guideline but it's not a strict rule. So, there are a number of limitations to these guidelines, so the first one is that, it is very specific to the idea of comparing and estimating magnitudes or quantities and once again, not everything in visualization is about comparing and estimating magnitudes. A lot is about it, but not everything. Another important aspect, is that sometime you may want to find a balance between accuracy and other important parameters. A very common one is scalability, there are situations where you want to visualize a lot of information at once and because of that, you just don't have enough space to use position or length as the main channel to represent a quantity. So, a typical example is a heat map, like the one that you see here. In this heat map we have lots of elements, lots of individual elements and magnitude is represented by color intensity. So now, color intensity is the lowest or one of the lowest channels in terms of accuracy in conveying quantitative information. Nonetheless, it's very useful in those situations where you need to represent a lot of data points at once, why? Well, because you don't need necessarily a lot of space in order to represent information with color. So once again, in these cases, keep in mind that you may want to find the balance between competing needs and a very common competing need, is the need of representing a lot of information, so the need to scale up. So, this concludes this part about accuracy.
