Now let's take a look at how to create color scales that are made to represent categorical information. So, what are desired properties of categorical color scales? In a way, they are similar, but also different from the properties that we wanted for quantitative color scales. So, the first one is uniformity. It's similar to what we said with the quantitative color scale, but also different. What do we mean by uniformity here? Well, we have to pick colors that represent different categories. As we will see in a moment, the best way to do that is to use different color use, color names. But when we use these colors, we don't want some of these colors to be brighter or lighter than other colors or more saturated and less saturated than other colors. Why? Well, the main idea is that if some colors are lighter or brighter or more saturated than others, they attract the attention more than the others. So, they create an imbalance between the classes. Unless this is desired, unless this is a desired property of your visualization, this shouldn't happen. The second one is discriminability. Think about it. What we are trying to do is to map categories in the data to different colors. So, we want to have as many colors as possible that can be distinguished perceptually, so that we can map as many categories as possible. But as we will see in a moment, there are not so many colors that we can actually very easily distinguish. But in general, what we want to achieve here is as much discriminability as possible because when we see colors in visualization, it has to be as easy as possible for the reader to distinguish between objects of a given color to objects of a different color. In other ways, if some objects have colors that are too similar, it's much, much easier to mix them up. So, now let me show you an example of a color scale that is meant to represent categorical information, but it's not uniform. There is one color that stands out more than the others. So, this is a map that is being created with the color palette that you see at the bottom right. As you can see, we can distinguish a number of different categories here, but there is one that is much brighter than the rest. Because of that, it stands out. So, once again, it is possible that in certain situations, you want to use this property. You do want some objects to stand out. But in general, if this is not desired, when you design or use a categorical color scale, you want to make sure that no color really stands out. So, this is an example with a different color scale, where all the colors have the same intensity lightness and the same saturation. In fact, you can see that no color really stands out, and we can perceive a lot of different categories. So, here there is another version of the map that is meant to convey the problem of discriminability. So, here we are using actually a color scale, a color palette where the color intensity and color saturation is uniform, is constant, but the number of available colors is reduced. Because of that, you can't really distinguish as many colors as in the previous image. So, look at the previous one and the next one. In this one, we can perceive a lot of different categories, whereas in this one, we can perceive only a few. So, how do we create categorical color scales. Once again, using the HCL space, the best way to do that is to choose a constant value of chroma and luminance and then sample uniformly across the hue value. So, let me show you once again through a demo with a color picker how these can be done. So, here we have the same color picker that we used and before. What I'm using here, I'm using a version where on the x-axis, we have hue, and on the y-axis, we have chroma. So, as you can see, colors up here are less vivid than colors down here. So, now, what do I do? Well, I want to span the use space as much as possible. So, I create a line that is as wide as possible here, and I want to draw it parallel to the chroma axis, so that I know that all these colors have exactly the same chroma. Since lightness can only be changed through the slider, I also know that the lightness, the color intensity is constant. So, now as you can see, I created a very nice color palette, which you can see here on the right side. I am spanning a number of different colors. They all have the same lightness value, and they also have the same chroma value. Let me reduce this here because otherwise it's too similar. So, let me show you how this color palette looks like when we visualize it on a map similar to what I've shown you before. So, this is a nice property of this color picker. So, here you can see that there are a number of different categories, and no one really stands out too much because all the colors have the same intensity, lightness, and the same chroma, which is the vividness, also called saturation.