When we talk about color, I like to start by quoting Edward Tufte, the famous statistician who wrote a number of incredibly beautiful and useful books on graphic design. So, this one comes from one of his best books called Envisioning Information. And the quote is, "Above all, do no harm." That's the most important rule when you're using color in visualization. So be careful because it's so easy to do things wrong. This is why I want to show you, first of all, a number of additional examples on what can go wrong with color when not used properly in visualization. So the first one is similar to the previous example. Again, color tend to be misused very often in maps. And here is another example. We have a map of Africa and there is some quantity that is mapped to different regions, but we can't really give an order to these quantities, because the colors that are used have no natural order. That's a very common problem. So, you try to map, some quantity to color, but the color that it's used doesn't have an inherent order. And because of that, it hinders appropriate perception of the values or decoding of the values from this map. That's another map, even more problematic than the previous one. So what the orders or designers tried to do here is to represent three different variables, three different pieces of information, using three different colors, and then map these colors to the regions or counties in the US map. So the first one is the percentage of high school students, the second one is the percentage of college graduates, and the third one is household income. And these colors are blended, so that hopefully, you can decode this information out of the map when you focus on some region. But as you can see, try to do that. If you focus on any of the regions that are displayed in the map, it's very hard to figure out what these three values are, okay? So, that's definitely problematic here. Here's a different example. This is complaints in New York City, and the volume of complaints at different times of the day. So here, there are different kinds of problems. So, a color here is used to represent different categories of complaints. But the problem is that, there are too many categories, and there are not enough colors to represent all these categories. And even the specific choice of colors is problematic because many of them look alike. So, there are many different purple colors, different reds, different yellows, and its not always immediate or easy to understand which is which. So, that's problematic. And an additional problem here is that these colors are so saturated that they create some sense of disorder and clutter in the visualization. So, there is also an aesthetic problem here. Here is another example that is somewhat similar but even more problematic. That's a pie chart with too many segments and too many colors. So in general, the idea here, the rule that we will see more details later on, if you want to visualize lots of categories with color, and there are too many categories, there's just not enough number of colors to represent all of them. But here, we have also the additional problem that some colors are re-used for different categories. So, for instance, green is used two times for two different categories. The same is true for orange. So, this is even more problematic. Here, we have a more subtle situation. So here in this graphics, colors is used to represent a difference between two years. So in these graphics, what is represented is information about diseases and how the percentage of diseases has changed over time between two different dates, and the color intensity is used to see, to represent the change between two years. But the problem, if you look at the legend, try to look at the legend for a moment, you'd see that some of these values are positive and some of these values are negative. So some of the changes between the two years are positive, and some of the changes are negative. But now, when you focus on the graphics again, you'd see that it's very hard to distinguish between diseases that had a positive change and diseases that had a negative change. So, we will see later on why this happens and what would be an appropriate color and coding for this situation. But I guess, you understand that there is a problem here. We can't really distinguish positive from negative and this is a problem.
