In the blog series "Subtleties of Color," written by Robert Simmon, he explains the importance of color for highlighting data and finding patterns and relationships. In data visualization, color is a crucial factor that affects how data is perceived and understood. For the sake of conveying data in a more precise manner, color cannot be picked randomly from a drop-down menu; often, that is how people choose colors, and it negatively affects data perception. For this, there are multiple color schemes, which suit different types of data in ways that emphasize data accordingly and ease readability.
The human eye and computers perceive color differently. On the one hand, humans perceive color in a non-linear and uneven way and are more sensitive to changes in lower lightness levels rather than high lightness levels and more sensitive to green light, then red light, and lastly, less sensitive to blue light. On the other hand, computer colors are linear and symmetrical and utilize systems that are not so "sympathetic" to the way human eyes perceive color. One of the main issues is that computers use the RGB color system (Red, Green, Blue), and due to how humans perceive changes in these three colors is different depending on the lightness, colors may not translate accurately.
Because of the way the human brain works, different palettes are more efficient in translating data and value differences more accurately than others. In that manner, sequential data is best represented by a color scheme that varies continuously and with even graduation. Divergent data, which is known as a dataset with two opposing values, such as the difference in temperature or the fluctuations of the stock market, is best represented with a divergent palette. A divergent palette is originated by the combination of two sequential palettes, joint in the center and expanding evenly in opposing, yet mirrored graduations. Last but not least, categorical data does not intend to represent proportional relationships but distinct categories. For this, a range of distinct, non-related colors, yet similar in contrast, is chosen; nevertheless, there is the counterpart of perceptual limitations, allowing a maximum of twelve, often fewer, colors per set.
Also, it is necessary to think of the audience. Sometimes palettes commonly utilized in scientific visualization may be confusing for a more general audience, therefore being better to use color palettes that are relatable to a broader public containing associations of color based on culture or nature. Color is crucial to better understand and read datasets; however, many factors affect how humans perceive color versus how screens translate these values. Color is very subjective, and
it is essential to be discerning when choosing a color palette to address the data and the audience better.