Now, I want to show you what happens when you are trying to represent additional information with these graphs. In particular, if you have additional attributes and you want to be able to introduce in the representation. Let's start with the bar chart. So, let's say that in a bar chart, now you want to represent an information coming from an additional categorical attribute. In this example that I have here, I'm using once again the vehicle collision data, and I'm representing in this case information about what kind of vehicle is involved in the accident, and what are the main causes for these accidents. So, we have vehicle type, main cause for the accident, and for each combination of these two categories, how many accidents or how many people injured we have. So, how do we represent these information with a bar chart? There are two main designs; the stacked bar chart, and the grouped bar chart. In a stacked bar chart, this information is represented as follows. So, we have as many bars as the number of categories that are included in the first categorical attribute and as many segments within each bar as the values that are in the other categorical attribute. So, in this case for instance, we have vehicle types represented by the three bars that you see in the stacked bar chart, and the main causes for collisions represented by the four colored, sorry, the five colored categories, okay? The five-colored main contributing factors for the collisions. In the grouped bar chart, we have exactly the same information but represented by a different configuration. In this case we have that the same bar graph is repeated multiple times for the number of categories that exist for the other categorical attribute. So, one categorical attribute is mapped to the bar chart and this bar chart is repeated as many times as the number of categories that you have in the other categorical attribute. So, what are the advantages and disadvantages of these two graphs? Well, the stacked bar graph is very good when your main question is regarding the proportion. So, if it's important for you to understand what is the proportion of values within each category, it's very good to communicate proportions, sometimes this is also called part-to-whole information. Whereas the grouped bar chart is better when the goal is to compare every single value one to another. So, now let me show you alternative designs for the line charts when we want to represent additional information on top of them. Once again, let's imagine that we have an additional categorical value that we want to be able to map using a line chart. So, in this case, we have, again, using the vehicle collision data, we have on the x-axis, time, month of the date, and on the y-axis we have number of collisions. But now we have an additional attribute, and this additional attribute here is the borough. So, the borough where these collisions happen. So, how is this accommodated in a line chart? In this first design, we see that we have separate lines, each one representing one borough. So, that's one possible design, when we want to represent an additional categorical attribute. This is very good because since we have the lines represented in the same space, it's easy to compare their values. The next variation is similar to the previous one, but every single line is represented in a separate line chart. So, it's a sort of series of line charts one next to the other, one above or below the other. So, that's another option to represent the same data. The last one I want to show you is the stacked version of the line chart, where the values across the categories are stacked one on top of the other. So, now these stacked area chart is good if the only question that you have is how does the proportion of these values change over time? But it's not very good for reading the individual values of these categories over time, and it's not good to compare the values of these categories over time. Why? Well, because the problem here is that since the values are stacked, the pattern that you see in the lines is affected by the shape of the line that is below the one that you are observing. This is clearly problematic because it's not the information that is represented in a stacked area chart like this one, cannot be consumed the same way you consume the information that is presented in a regular line chart. As I said before, because the pattern doesn't really represent the values that are associated to each line chart.
